Keywords: Affective Computing, Brain-based Information Communications, Human-Computer Interaction
Abstract: Electroencephalography (EEG)-based emotion recognition has advanced the field in affective computing and has enabled applications in human-computer interactions. Despite significant progress has been made in decoding emotion using supervised machine-learning methods, few studies applied data-driven, unsupervised approaches to explore the underlying EEG dynamics during an emotion experiment and examine how such dynamics correlate with subjective reports of emotion. This study employs the adaptive mixture independent component analysis (AMICA), an unsupervised approach, to EEG data from the DEAP dataset where 32 subjects watched emotional videos. Empirical results showed that AMICA could learn distinct models that separated EEG date collected in the emotion experiment. The identified changes in EEG patterns were weakly-correlated with the four reported emotion scales, indicating the underlying EEG dynamics partially reflected the emotional activities as well as the emotion-irrelevant brain dynamics. Further, the correlations between EEG dynamics and individuals' subjective emotional ratings were significantly higher than those between the EEG and the average ratings from online raters. Finally, building an emotion-decoding model based on the EEG dynamics revealed a significantly better classification performance for valence ratings compared to arousal. This study demonstrated the use of AMICA in characterizing the EEG dynamics in emotion experiments and provided insight into the relationship between EEG and the reported emotional experiences. The unsupervised learning approach can be applied to studying emotion and other confounding factors such as emotion irrelevant EEG artifacts, thereby improving the performance of emotion decoding for EEG-based affective computing.

Keywords: Technology Assessment, Model-based Systems Engineering, Systems Medicine
Abstract: The accurate measurement of brain activity by Brain-Machine-Interfaces (BMI) and closed-loop Deep Brain Stimulators (DBS) is one of the most important steps in communicating between the brain and subsequent processing blocks. In conventional chest-mounted systems, frequently used in DBS, a significant amount of artifact can be induced in the sensing interface, often as a common-mode signal applied between the case and the sensing electrodes. Attenuating this common-mode signal can be a serious challenge in these systems due to finite common-mode-rejection-ratio (CMRR) capability in the interface. Emerging BMI and DBS devices are being developed which can mount on the skull. Mounting the system on the cranial region can potentially suppress these induced physiological signals by limiting the artifact amplitude. In this study, we model the effect of artifacts by focusing on cardiac activity, using a current- source dipole model in a torso-shaped volume conductor. Performing finite element simulation with the different DBS architectures, we estimate the ECG common mode artifacts for several device architectures. Using this model helps define the overall requirements for the total system CMRR to maintain resolution of brain activity. The results of the simulations estimate that the cardiac artifacts for skull-mounted systems will have a significantly lower effect than non-cranial systems that include the pectoral region. It is expected that with a pectoral mounted device, a minimum of 60-80 dB CMRR is required to suppress the ECG artifact, depending on device placement relative to the cardiac dipole, while in cranially mounted devices, a 0 dB CMRR is sufficient, in the worst-case scenario. In addition, the model suggests existing commercial devices could optimize performance with a right-hand side placement. The methods used for estimating cardiac artifacts can be extended to other sources such as motion/muscle sources. The susceptibility of the device to artifacts has significant implications for the practical translation of closed-loop DBS and BMI, including the choice of biomarkers, the system design requirements, and the surgical placement of the device relative to artifact sources.

Keywords: Human-Computer Interaction, Human-Machine Interface, Brain-based Information Communications
Abstract: Brain-computer interfaces (BCIs) based on code-modulated visual evoked potentials (c-VEPs) hold promise to serve as a fast and reliable hands-free communication tool for people with severe disabilities. A c-VEP BCI application presents flickering target objects (e.g. letters of a keyboard) coded with different time-lags of a code pattern. Template matching methods are used to identify the target of interest. Unfortunately, this approach requires a training session, in which several trials of EEG data are recorded and analysed. Long training sessions are necessary to ensure good signal-to-noise ratios. For the user, these training sessions may be tedious. Especially for patients, who may use the system on a daily basis e.g. for communication, alternative approaches are desirable. This paper investigates the feasibility of session-to-session transfer of EEG templates for c-VEP BCIs, where templates recorded in a previous session are used, so the application could be used instantly. Ten healthy participants went through training and copy-spelling tasks in two experimental sessions (they were scheduled two weeks apart). In the second session, the templates recorded in the first session were used. Eight participants yielded good results with the session-to-session transfer approach with accuracies of 97.1 per cent and information transfer rates of 85.7 bit/min on average. For these participants, the results were not significantly different from the values achieved using the standard approach (training in the same session). For two participants, however, the system was not controllable with the priorly recorded templates. The results demonstrate that for most users daily recalibration is not required.

Keywords: Brain-based Information Communications, Human-Machine Interface, Human-Computer Interaction
Abstract: Asynchronous Brain Computer Interfaces (BCI) allow system activation at free will. This scenario would be desirable for potential users such as patients with motor disabilities, however, there are still several limitations such as long calibration times and overall performance. Different approaches to improve these systems explore diverse linear electroencephalography (EEG) features in order to describe cognitive states. Alternatively, the present study proposes the use of recurrence quantification analysis (RQA), which provides indices about the dynamical behaviour of non-linear systems. We evaluated the accuracy of continuous detection of mental calculation and idle state using RQA features from EEG recordings. Among the different classification methods that were tested, random forest (RF) classifier resulted in the best performance. From 35 selected RQA features, 12 healthy subjects achieved mean accuracy of 0.924 ± 0.048, whereas with 150 selected features mean accuracy was 0.977 ± 0.020. These results suggest that RQA features are adequate for continuous task episode detection in asynchronous BCI. However, a restriction in the number of these features, conceived for improving computational costs and time, does impact detection performance.

Keywords: Human Factors, Human Performance Modeling, Human-Machine Interface
Abstract: Brain-computer interfaces (BCI) translate brain activity into an action that is carried out by a computer or robotic device. Motor-imagery BCIs (MI-BCI) rely on the user’s imagination of bodily movements, however not all users can generate the brain activity needed to control MI-BCI. This difference in MI-BCI performance among novice users could be due to their cognitive abilities. In this study, we investigated the impact of spatial abilities and visuospatial memory on MI-BCI performance. Fifty-four novice users participated in a MI-BCI task and two cognitive tests. The impact of spatial abilities and visuospatial memory on BCI task error rate in three feedback sessions was measured. Our results showed that spatial abilities, as measured by the Mental Rotation Test, were not related to MI-BCI performance, however visuospatial memory, assessed by the design organization test, was higher in high aptitude users. Our findings can contribute to optimization of MI-BCI training paradigms through participant screening and cognitive skill training.

Keywords: Evolutionary Computation, Optimization
Abstract: In machine learning field, feature selection can be treated as a bi-objective optimization problem. It is reported that a decomposition-based evolutionary multi-objective optimization algorithm (i.e., MOEA/D-STAT) has good diversity performance when coping with feature selection. However, feature selection is also a time-consuming problem considering a large dataset it involves. The computation time can be easily reduced by introducing the parallelization into MOEA/D-STAT, thanks to the decomposition idea of MOEA/D. To the best of our knowledge, this is the first attempt to implement the parallelization of MOEA/D-STAT for feature selection. In this paper, we consider both master-slave models and island models, which are two different approaches of parallelization. In the master-slave models, different offspring assignment mechanisms are considered. In the island models, different island size specification mechanisms are examined. Our experimental results show that the master-slave models can achieve higher speedup and better performance than the island models.

Keywords: Evolutionary Computation, Optimization
Abstract: Genetic algorithm (GA) is a widely used probabilistic search optimization algorithm. In the GA, selection is an important operator to guarantee the quality of solution. Therefore, the behavior of selection operator makes a great effect of the performance of the algorithm. This paper designs a new and efficient selection operator for GA base on the idea of promotion competition. This operator simulates the rule and process of promotion competition to protect the well perform chromosomes and eliminates poor chromosomes. This is a fundamental but significant research issue in GA that may be adopted into any existing GA variants to replace any other selection operators. We design four types of experiments to comprehensively verify the behavior of the proposed promotion selection operator, by comparing it with five other existing and commonly used selection operators. The results show that promotion selection operator has a general good performance in enhancing GA in terms of solution quality, convergence speed, and running time.

Keywords: Evolutionary Computation, Optimization
Abstract: Classical benchmark problems utilize multiple transformation techniques to increase optimization difficulty, e.g., shift for anti centering effect and rotation for anti dimension sensitivity. Despite testing the transformation invariance, however, such operations do not really change the landscape’s “shape”, but rather than change the “view point”. For instance, after rotated, ill conditional problems are turned around in terms of orientation but still keep proportional components, which, to some extent, does not create much obstacle in optimization. In this paper, inspired from image processing, we investigate a novel graphic conformal mapping transformation on benchmark problems to deform the function shape. The bending operation does not alter the function basic properties, e.g., a unimodal function can almost maintain its unimodality after bent, but can modify the shape of interested area in the search space. Experiments indicate the same optimizer spends more search budget and encounter more failures on the conformal bent functions than the rotated version. Several parameters of the proposed function are also analyzed to reveal performance sensitivity of the evolutionary algorithms.

Keywords: Evolutionary Computation, Optimization, Heuristic Algorithms
Abstract: Many-objective optimization problems (MaOPs) are multi-objective problems that have four or more objectives. MaOPs face significant challenges because of search inefficiency, computational cost, decision making, and visualization. Most MaOP systems use variants of non-dominated sorting (Pareto ranking). However, Pareto dominance is ineffective when the number of objectives exceeds four. In this research, we explore different strategies for solving MaOPs. We use Hornby's Age-Layered Population Structure (ALPS) evolutionary algorithm in order to mitigate premature convergence and improve results. Instead of Pareto ranking, we use the many-objective evaluation strategy called sum of ranks (SR). SR is more appropriate than Pareto dominance for problems that require a majority of objectives to be optimized. We introduce and compare different objective reduction methods for ALPS, including random and correlated objective reduction. Because hypervolume and IGD performance measurements are not necessarily suitable to SR strategies, we introduce a new minimum distance measurement. Results show that different strategies are suitable for different problems, and depend strongly on the performance measure being used. Random objective reduction was the least effective strategy, while correlated reduction was more successful. The research shows that the ALPS framework with objective reduction is a promising framework for MaOPs.

Keywords: Evolutionary Computation, Optimization, Heuristic Algorithms
Abstract: In the last decades, evolutionary algorithms (EAs) have demonstrated strong search capabilities in solving multi-objective optimization problems (MOPs). To improve the search performance of EAs, as problems seldom exist in isolation, transferring knowledge from related problems have attracted considerable attentions in recent years. In this paper, we present a preliminary study to enhance existing evolutionary algorithms (MOEAs) by transferring knowledge from the process of solving the single objectives involved in a given MOP of interest. As the single objectives are the objectives of the MOP, they naturally share great similarity with the given MOP, which thus could yield useful traits for enhancing the problem-solving of the MOP. To the best of our knowledge, this work severs as the first attempt to improve evolutionary multi-objective optimization via transferring knowledge from single objective problems. To evaluate the performance of the proposed method, empirical studies using a popular MOEA, i.e., NSGAII, on commonly used multi-objective benchmarks are conducted. The obtained results confirmed the efficacy of the proposed method in terms of both convergence speed and solution quality.

Keywords: Biometric Systems and Bioinformatics, Image Processing/Pattern Recognition, Neural Networks and their Applications
Abstract: Automatic segmentation of retinal vessels is an important step in fundoscopic image analysis. Recently, convolutional-neural-network-based methods have been widely explored in this vision task. However, the local fixed receptive field makes network unable to collect global information and adapt to scale variation of retinal vessels. In this paper, we propose a novel RNA-Net which can capture nonlocal context dependencies across the inputs and extract multi-scale features for segmentation task. Firstly, we build a Residual Nonlocal Attention (RNA) Module, which can guide the network to pay more attention to task-related regions of the whole feature map. Secondly, to better capture the morphological characteristics of natural blood vessels, Pyramid Pooling Module (PPM) is added to capture features at multiple scales. Experimental results on two public datasets DRIVE and STARE clearly demonstrate that our method outperforms the current state-of-the-art approaches.

Keywords: Biometric Systems and Bioinformatics, Neural Networks and their Applications, Image Processing/Pattern Recognition
Abstract: 3D ultrasound imaging shows great promise for scoliosis diagnosis thanks to its low-costing, radiation-free and real-time characteristics. The key to accessing scoliosis by ultrasound imaging is to accurately segment the bone area and measure the scoliosis degree based on the symmetry of the bone features. The ultrasound images tend to contain many speckles and regular occlusion noise which is difficult, tedious and time-consuming for experts to find out the bony feature. In this paper, we propose a robust bone feature segmentation method based on the U-net structure for ultrasound spine Volume Projection Imaging (VPI) images. The proposed segmentation method introduces a total variance loss to reduce the sensitivity of the model to small-scale and regular occlusion noise. The proposed model improves 2.3% of Dice score and 1% of AUC score as compared with the u-net model and shows high robustness to speckle and regular occlusion noise.

Keywords: Image Processing/Pattern Recognition
Abstract: The performance of many methods of MRI-based computer-aided diagnosis of breast disease rely on accurate delineation of the breast boundary. This problem has been knownto be a challenging one on the account of the complex compositionof breast tissue and its extensive inter-subject variability. To address this problem, this paper introduces a new approach to whole-breast segmentation which, as opposed to many existingsolutions, can operate in the absence of any prior informationon the patient-specific breast anatomy. The proposed algorithmtakes advantage of Dijkstra’s procedure which allows accurately tracking the boundary between the pectoralis muscle and breast tissue as well as between the breast and its background. Theperformance of the proposed method has been tested on in vivo MRI volumes and quantified in terms of several performancemetrics. The experimental results demonstrate consistent and stable performance of the proposed algorithm in terms of itsaccuracy and robustness to imaging artefacts

Keywords: Image Processing/Pattern Recognition, Neural Networks and their Applications, Machine Learning
Abstract: Recently, the automatic diagnosis of Turner syndrome (TS) has been paid more attention. However, existing methods relied on handcrafted image features. Therefore, we propose a TS classification method using unsupervised feature learning. Specifically, first, the TS facial images are preprocessed including aligning faces, facial area recognition and processing of image intensities. Second, pre-trained convolution filters are obtained by K-means based on image patches from TS facial images, which are used in a convolutional neural network (CNN); then, multiple recursive neural networks are applied to process the feature maps from the CNN to generate image features. Finally, with the extracted features, support vector machine is trained to classify TS facial images. The results demonstrate the proposed method is more effective for the classification of TS facial images, which achieves the highest accuracy of 84.95%.

Keywords: Machine Learning, Neural Networks and their Applications, Image Processing/Pattern Recognition
Abstract: Analyzing the chest CT images are used as an alternative to COVID-19 tests as we are in very short supply of detecting test kits. Automating the process of analyzing can save great amount of time and energy. In this paper a bayesian ensembling framework is proposed for automatic detection of COVID-19 cases using the chest CT scans. Data augmentation is applied to increase the size and quality of training data available. Transfer learning is utilized to extract the features. The extracted features are used to train the three different bayesian classifiers. The uncertainty of the neural network predictions is estimated by anchored, unconstrained and regularized bayesian ensembling methods. The reliability of the predictions is delineated. The epistemic and aleatoric uncertainties are estimated and different bayesian classifiers are compared. We use a small dataset containing only 275 CT images of positive COVID-19 cases. The results sounds promising and they can be improved in the future, as the performance of deep neural networks are reliant to big datasets.

Keywords: Hybrid models of NN, Agent-Based Modeling
Abstract: Federated Learning (FL), a distributed version of Deep Learning (DL), was introduced to tackle the problem of user privacy and huge bandwidth requirements in sending the user data to the company servers that run DL models. FL enables on-device training of the models. Most FL approaches are entirely centralized and suffer from inherent limitations such as single node failure and channel bandwidth bottlenecks. To circumvent these issues, we present an approach to decentralize FL using mobile agents coupled with the Federated Averaging (FedAvg) algorithm. A hybrid model that combines both centralized and decentralized approaches has also been presented. Results obtained by running the model on different network topologies indicate that the hybrid version proves to be the better option for an FL implementation.

Keywords: Hybrid models of NN, Neural Networks and their Applications, Computational Intelligence
Abstract: Radiologists are required to write free paper text reports for breast screenings in order to assign cancer diagnoses in a later step. The current procedure requires considerable time and needs efficiency. In this paper, to streamline the writing process and keep up with the specific vocabulary, a word prediction tool using neural language models was developed. Consequently, challenges as different languages (English, Dutch), small data sizes and low computational power have been overcome by introducing a novel English-Dutch Radiology Language Modelling process. After defining model architectures, the process involves data preparation, bilevel hyperparameters optimization, configuration transfer and evaluation. The model is able to improve the current workflow and successfully meet the computational constraints, based on both an intrinsic and extrinsic evaluation. Given its flexibility, the model opens the door for future research involving other languages and also an extensive set of real-world applications.

Keywords: Hybrid models of NN, Neural Networks and their Applications, Machine Learning
Abstract: Traffic flow prediction is an important foundation for intelligent transportation systems. The traffic data are generated from a traffic network and evolved dynamically. So spatio-temporal relation exploration plays a support role on traffic data analysis. Most researches focus on spatio-temporal information fusion through a convolution operation. To the best of our knowledge, this is the first work to suggest that it is necessary to distinguish the two aspects of spatial correlations and propose the two types of spatial graphs, named as geographic graph and semantic graph. Then two novel stereo convolutions with irregular acceptive fields are proposed. The geographic-semantic-temporal contexts are dynamically jointly captured through performing the proposed convolutions on graph sequences. We propose a geographic-semantic-temporal graph convolutional network (GST-GCN) model that combines our graph convolutions and GRU units hierarchically in a unified end-to-end network. The experiment results on the Caltrans Performance Measurement System (PeMS) dataset show that our proposed model significantly outperforms other popular spatio-temporal deep learning models and suggest the effectiveness to explore geographic-semantic-temporal dependencies on deep learning models for traffic flow prediction.

Keywords: Machine Learning, Hybrid models of NN, Intelligent Internet Systems
Abstract: Twitter is a popular social networking platform. While people enjoy the news and anecdotes on Twitter, there are also lots of rumors, which have a negative impact on users and can compromise social order. Among these rumors, many of them are written in Cantonese. At present, the research of English rumor detection is relatively comprehensive, but Cantonese rumors are rarely studied, which brings great challenges to the detection of Cantonese rumors on Twitter. Firstly, there is no available benchmark dataset of Cantonese rumors. Secondly, it is difficult to completely extract the features of rumors. Thirdly, the classical detection approaches are not effective in detecting Cantonese rumors. In this paper, we collected and annotated Cantonese rumors on Twitter and obtained a relatively complete Cantonese rumor dataset. Next, 27 statistical features, involving four categories (user, content, propagation, and comment-based), are extracted to distinguish rumors and non-rumors in Cantonese. Seven of these features are newly proposed in this paper. Then, a novel deep learning model called BLA (namely BERT-based Bi-LSTM network with Attention mechanism) is built for Cantonese rumors detection on Twitter. BLA takes advantage of both statistical and semantic features to effectively detect Cantonese rumors. The experimental results show that the BLA model outperforms other detection models in Cantonese rumor detection.

Keywords: Machine Learning, Hybrid models of NN, Neural Networks and their Applications
Abstract: Weather prediction is one of the hot topics in artificial intelligence. In this paper, three new temperature prediction models based on historical data are proposed for two important meteorological indexes, the maximum temperature and the minimum temperature. The first model is to construct SVM model to predict the residual error of LSTM model, then add the prediction results of the two models to get the final prediction result. The second model is to use empirical mode decomposition (EMD) to decompose the original data, then use the combination forecasting model to predict the subsequences, and finally summarize the prediction results. The third model is to combine the advantages of the first and second models. First, EMD is used to decompose the original sequence, and the first model is used to predict each subsequence. Finally, the predicted values of all subsequences are superimposed to obtain the final predicted value. Based on the temperature data of Washington and Los Angeles, the three models are tested and analyzed in this paper. The experimental results show that the third model proposed in this paper, which is based on EMD and residual prediction SVM-LSTM model, has better prediction accuracy than other models.

Keywords: Machine Learning, Expert and Knowledge-based Systems, Hybrid models of NN
Abstract: A bug tracking system continuously monitors the status of a software environment, like an Operating System (OS) or user applications. Whenever it detects an anomaly situation it generates a bug report and sends it out to the software developer or maintenance center. However, the newly reported bug can be an already existing issue that was reported earlier and may have a solution in the master report repository at the developer side. Such instances may occur repeatedly in an overwhelming number. This poses a big challenge to the developer. Thus, early detection of duplicate bug reports has become an extremely important task. This work proposes a double-tier approach using clustering and classification, whereby it exploits Latent Dirichlet Allocation (LDA) for topic-based clustering, multimodal text representation using Word2Vec (W2V), FastText (FT), and Global Vectors for Word Representation (GloVe), and text similarity measure fusing Cosine and Euclidean measures. The proposed model is tested on the Eclipse dataset consisting of over 80,000 bug reports, which is the amalgamation of both master and duplicate reports. This work only considers the description of the reports for detecting duplicates. The experimental results show that the proposed two-tier model achieves a recall rate of 67% for Top-N recommendations in 3 times faster computation than the conventional one-on-one classification model.

Keywords: Machine Learning, Heuristic Algorithms
Abstract: Dynamic regressor selection (DRS) systems work by selecting the most competent regressors from an ensemble to predict the target value of a query pattern. This competence is calculated using the performance of the regressors in a local region of the feature space around the query pattern that is called the region of competence. However, choosing the correct measure to calculate the level of competence of the regressors is a hard task. In this paper, we propose a new method to DRS that selects the best competence measure for a given dataset. To validate our method, we perform a set of comprehensive experiments on 15 regression datasets. The proposed method can operate in three different fashions: (I) selection of the most competent regressor; (ii) combination of all regressors from the ensemble; and (iii) selection of a subset composed of the most competent ones and combine them. The proposals are compared against DRS algorithms, individual regressors, and static systems that use the Mean and the Median as fusion strategy. The results show that the proposed method, which chooses a different competence measure per task, outperforms literature methods.

Keywords: Machine Learning, Image Processing/Pattern Recognition
Abstract: Learning semantically disentangled representations is important for various computer vision tasks, such as image generation and classification. Although it is possible to learn an effective representation in supervised settings, there are problems requiring enormous effort focused in data collection and labeling, and the difficulty in labeling continuously changing events like facial expressions is significant. In this paper, we propose a method for separating the latent representation of facial images into identity factors and facial expression factors using the variational autoencoder (VAE) framework. In our method, we only use subject labels to control training, and we do not use information attached to facial expressions like emotional labels. The separation between extracted facial expression factors and identity features is very useful for controlling image generation and for classifying facial expressions. Using this latent representation, we also suggest a new approach for facial expression recognition with a simple clustering method, which is based on Euclidean distance. Our classification method dramatically reduces the cost of labeling. The experimental results show that our method successfully disentangles the representation of facial images and separates the latent representation into identity and facial expression factors. Moreover, in a facial expression recognition task, our approach shows advantages over the baseline method without supervision.

Keywords: Machine Learning, Image Processing/Pattern Recognition, Machine Vision
Abstract: Facial expression is an important facial semantics on visual aspect. The facial expressions synthesis has a wide range of applications in human-computer interaction and virtual reality. In recent years, image synthesis base on generative adversarial networks(GANs) is developing rapidly. In the image-to-image translation work, we propose a new facial expression generation method base on the idea of conditional GANs and realize the optimization of the generated results. The main work of this paper includes: Editable facial lines map is utilized as a constraint, combining with neutral face images as inputs of generator, so that a variety of facial expression images can be generated by editing the constraints. Correntropy loss of feature matching is added, which is used to measure the intermediate representation between the real images and the generated images by improving the adversarial loss. Consequently, the generated facial expressions can be more realistic. Base on the ideas above, the proposed method needs only one generator to generate different realistic facial images with various expressions.

Keywords: Machine Learning, Industry 4.0
Abstract: China's industry urgently needs to be transformed from the development patterns driven by traditional production factor to achieving industrial ecological transformation (IET). The IET is influenced by diversified factors including resource input, allocation and flow, environmental regulations and technological innovations in different situation. Revealing the complex influence mechanisms between IET and its influence factors is necessary for effectively analyzing, evaluating and improving the performance of IET. A three stages machine learning method including learning, verification and generalization based on dominance-based rough set approach is presented to extract the influential relationships between the IET and its contextual influence factors. The proposed method excavates and learns the historical panel data of China's 30 provinces, and the cross-validation is conducted to produce a set of highly credible "If-Then" decision rules to generalize the synergistic influential relationships and intensities in IET. The results show that China's investment strength, resource allocation efficiency, command controlled and economic incentive environmental regulations are determinants to enhance the performance of IET, which helps to select the optimal transformation patterns by taking the historical development characteristics as lessons.

Keywords: Neural Networks and their Applications, Machine Learning, Evolutionary Computation
Abstract: Although convolutional neural networks (CNNs) have significantly evolved and demonstrated outstanding performance, their　uninterpretable nature is still considered to be a major problem. In this study, we take a closer look at CNN interpretability and propose a new method called Evolutionary Generative Contribution Mappings (EGCM). In EGCM, CNN models incorporate both a classification mechanism and an interpreting mechanism in an end-to-end training process. Specifically, the network generates the class contribution maps, which indicate the discriminative regions for the model to identify a specific class. Additionally, these maps can be directly used for classification tasks; all that is needed is a global average pooling and a softmax function. The network is represented by a directed acyclic graph and optimized using a genetic algorithm. Architecture search enables EGCM to deliver reasonable classification performance while maintaining high interpretability. We apply the EGCM framework on several datasets and empirically demonstrate that the EGCM not only achieves excellent classification performance but also maintains high interpretability.

Keywords: Neural Networks and their Applications, Machine Learning, Image Processing/Pattern Recognition
Abstract: In this paper, we proposed a new clothing brand prediction method which is rooted on a dense-block based deep convolutional neural network for brand logo detection and recognition. To learn convolutional neural networks deeper and more accurately, we adopted dense blocks into deep convolutional neural networks to make connections between layers shorter. In this work, we propose several dense-block based designs to improve clothing brand logo detection and recognition accuracies. We also constructed a new large-scale clothing brand and price (CBP) dataset and its subset, called clothing brand logo (CBL) dataset with the brand attribute and logo information to carry out this task. To lower proposed framework complexity, two pixel search steps for the bounding box movement are implemented in the training procedure. In the experiment, we show our search reduced model can outperform several state-of-the-art methods and attain good performance.

Keywords: Neural Networks and their Applications, Machine Learning, Multimedia Computation
Abstract: Speech is converted to digital signals using speech coding for efficient transmission. However, this often lowers the quality and bandwidth of speech. This paper explores the application of convolutional neural networks for Artificial Bandwidth Expansion (ABE) and speech enhancement on coded speech, particularly Adaptive Multi-Rate (AMR) used in 2G cellular phone calls. In this paper, we introduce AMRConvNet: a convolutional neural network that performs ABE and speech enhancement on speech encoded with AMR. The model operates directly on the time-domain for both input and output speech but optimizes using combined time-domain reconstruction loss and frequency-domain perceptual loss. AMRConvNet resulted in an average improvement of 0.425 Mean Opinion Score – Listening Quality Objective (MOS-LQO) points for AMR bitrate of 4.75k, and 0.073 MOS-LQO points for AMR bitrate of 12.2k. AMRConvNet also showed robustness in AMR bitrate inputs. Finally, an ablation test showed that our combined time-domain and frequency-domain loss leads to slightly higher MOS-LQO and faster training convergence than using either loss alone.

Keywords: Neural Networks and their Applications, Machine Learning, Optimization
Abstract: The prediction of long-term user group activity and cluster of activities around a subject in social networks is a very challenging task. In this paper, we propose a novel temporal neural network framework that tracks user engagement and activity associated with particular subjects (e.g. CVE ID) across online platforms. The framework is able to simulate which user will do what activity and at what time. Furthermore, this framework captures groups of users reacting to an event. It also captures responses to an event on a platform and the influence on the event on activity on other platforms over time.

The proposed framework aims to predict future user activity related to specific subjects across platforms. The framework also illustrates the importance of the influence of activities that occur on a platform on other platforms to predict user activity for particular events. The learned model can do simulations in a timely manner.

We evaluated our user group activity prediction method on the CVE (Common Vulnerabilities and Exposures) related user groups (software vulnerability) using 3 public online social network datasets: Github, Reddit, and Twitter. Groups of users who work on a particular CVE ID are identified. Each user group records all users' activities related to a CVE ID. The 3 datasets from Github, Reddit, and Twitter contain more than 490,000 cross platform activities related to over 20,000 user groups (CVE IDs) from more than 50,000 users. Compared to the proposed baseline, our simulation method is better in both predictions of total activity volume over time and activity associated with individual CVE ID.

Keywords: Neural Networks and their Applications, Machine Vision, Image Processing/Pattern Recognition
Abstract: Traditional convolutional neural network classification models often only focus on the most distinguishing feature regions of the image and ignore the weaker feature regions. However, the image position distribution of plant diseases is very uneven. If we use convolutional neural network for plant disease recognition, there will be insufficient feature response, which will cause recognition errors. Aiming at such problems, we have designed a deep complementary feature classification network. First, the network uses DeepLabv3+ and Conditional Random Field (CRF) to generate disease part detection frames in a weakly supervised manner and combines semantic segmentation to extract disease object instances. Then we designed Complementary Feature Part Generation Models. Finally, it uses a bidirectional Gated Recurrent Unit (Bi-GRU) to perform the classification and recognition of the complementary features described above. We performed experiments on the PlantVillage dataset. The experimental results show that the proposed network recognition accuracy is 99.21%, which is 4.2% higher than the baseline model xception-65 used. We also performed experiments on the grape disease data set that we created. The accuracy of the proposed network recognition is 93.46%, which is 7.2% higher than the baseline model xception-65. In addition, compared with the better algorithms for plant disease identification in recent years, the accuracy performance has also been improved.

Keywords: Optimization, Heuristic Algorithms
Abstract: This paper suggests the merge of the positive features of two data-driven control techniques, namely second order Active Disturbance Rejection Control (ADRС) and Virtual Reference Feedback Tuning (VRFТ). The efficiency of this combination, referred to as ADRС-VRFТ, is tested by comparison with data-driven ADRС through experimental results on nonlinear twin rotor aerodynamic system (ТRAЅ) laboratory equipment. The parameters of ADRС-VRFТ algorithm are obtained in a model-free way and the parameters of ADRС algorithm are obtained in a model-based one making use of the nonlinear first principles mathematical model of ТRAЅ

Keywords: Image Processing/Pattern Recognition, Machine Learning, Computational Intelligence
Abstract: The automatic segmentation of medical images represents a research domain of high interest. This paper proposes an automatic procedure for the detection and segmentation of gliomas from multi-spectral MRI data. The procedure is based on a machine learning approach: it uses ensembles of binary decision trees trained to distinguish pixels belonging to gliomas to those that represent normal tissues. The classification employs 100 computed features beside the four observed ones, including morphological, gradients and Gabor wavelet features. The output of the decision ensemble is fed to morphological and structural post-processing, which regularize the shape of the detected tumors and improve the segmentation quality. The proposed procedure was evaluated using the BraTS 2015 train data, both the high-grade (HG) and the low-grade (LG) glioma records. The highest overall Dice scores achieved were 86.5% for HG and 84.6% for LG glioma volumes.

Keywords: Machine Learning, Computational Intelligence, Neural Networks and their Applications
Abstract: Most multi-layer neural networks used in deep learning utilize rectified linear neurons. In our previous papers, we showed that if we want to use the exact same activation function for all the neurons, then the rectified linear function is indeed a reasonable choice. However, preliminary analysis shows that for some applications, it is more advantageous to use different activation functions for different neurons – i.e., select a family of activation functions instead, and select the parameters of activation functions of different neurons during training. Specifically, this was shown for a special family of squashing functions that contain rectified linear neurons as a particular case. In this paper, we explain the empirical success of squashing functions by showing that the formulas describing this family follow from natural symmetry requirements.

Keywords: Fuzzy Systems and their applications, Expert and Knowledge-based Systems, Image Processing/Pattern Recognition
Abstract: A system called the Computer-Assisted Surgical Trainer (CAST) that provides force, visual, and audio guidance with objective assessments for laparoscopic surgery skills training in a non-patient training setup was developed. To support the guidance and assessment of trainees, given an object transfer task, a fuzzy reasoning system is proposed to estimate states of objects as well as the completion of basic actions. A peg transfer task is implemented as a target application. Based on a modeling approach of this peg transfer task, fuzzy production rules are defined and a reasoning process is designed. The experimental test results show the effectiveness of the proposed fuzzy reasoning system.

Keywords: Human Factors, Assistive Technology, Human-Machine Interface
Abstract: Driving automation leads drivers to be disengaged by dynamic driving tasks, but the need for the driver to monitor traffic environment is existed in a partial automation. Monotonous road situations can produce drowsiness, resulting in poor driving performance. Conversation in automated vehicles is expected to mitigate fatigued driving. However, few studies have examined effects of conversation. This paper reports a driving simulator study describing impacts of passenger conversation on driver fatigue and driving performance regarding the relationship between a driver and a passenger. Participants were categorized into two groups: a passenger’s friends (ConvR) and people who had not met the passenger before (ConvNR) to examine whether the acquaintance with the passenger affects driver behaviour, driving performance, fatigue and workload. The results showed that drivers in ConvNR felt lesser sleepy and found it easier to drive than the drivers in ConvR, but no difference of driver behaviour and driving performance was observed between two groups. In addition, regardless of the relationship, making casual conversation led decreased fatigue and workload in comparison with solo driving without conversation. The practical implication for further conversation methods to regard driver fatigue and driving performance in the automated vehicle is discussed.

Keywords: Human Factors, Wearable Computing, Assistive Technology
Abstract: In recent years, driver's delayed recognition has caused many traffic accidents. Cognitive effort decreases awareness and delays the driver’s attention on the surrounding environment. Conventionally, the degree of cognitive load on a driver, namely, the driving workload, is estimated from the steering pattern of the steering wheel. Direct measurements of the body movements operating the vehicle might more easily and accurately detect the small changes caused by driving workload than conventional methods. Therefore, we investigate the effect of cognitive load on the steering operation and body behavior of drivers, and verify the applicability of our approach to driving-workload estimation. The physical behavior refers to the behavior of the hands operating the steering wheel. From the acceleration of the hands, we derive an index of the driving workload. The proposed method was experimentally evaluated on seven subjects performing a dual task. The estimation accuracy of the proposed method at least matched that of the conventional steering-entropy method.

Keywords: Human-Machine Cooperation and Systems, Human Factors
Abstract: Understanding a driver's behavior in a steering task is essential to the development of haptic guidance systems. This paper aims to predict driver torque control, especially when haptic guidance is part of haptic feedback. A new cybernetic driver model with an improved neuromuscular system is proposed and identified. It is assumed that the driver converts a target steering-wheel angle into torque by both indirect and direct control. Indirect control refers to the adaptation of the parameters of an internal model of steering compliance as perceived by the driver. Direct control accounts for the driver's corrective action through direct haptic feedback. The parameters of the model were identified with data collected from experiments conducted with a driving simulator. The results of identification were satisfactory and led to good representation of the driver's action, with or without haptic guidance. The model accurately predicted driver torque output. It can be used to study driver adaptation to haptic guidance systems.

Keywords: Human-Machine Cooperation and Systems, Human Factors, Assistive Technology
Abstract: Automatic emergency steering maneuvers can be used to avoid more obstacles than emergency braking alone. While a steer-by-wire system can decouple the driver who might act as a disturbance during the emergency steering maneuver, the alternative in which the steering wheel remains coupled can enable the driver to cover for automation faults and conform to regulations that require the driver to retain control authority. In this paper we present results from a driving simulator study with 48 participants in which we tested the performance of three emergency steering intervention schemes. In the first scheme, the driver was decoupled and the automation system had full control over the vehicle. In the second and third schemes, the driver was coupled and the automation system was either given a high impedance or a low impedance. Two types of unexpected automation faults were also simulated. Results showed that a high impedance automation system results in significantly fewer collisions during intended steering interventions but significantly higher collisions during automation faults when compared to a low impedance automation system. Moreover, decoupling the driver did not seem to significantly influence the time required to hand back control to the driver. When coupled, drivers were able to cover for a faulty automation system and avoid obstacles to a certain degree, though differences by condition were significant for only one type of automation fault.

Keywords: Human-Machine Cooperation and Systems, Human-Machine Interface, Human Factors
Abstract: A concern has been raised regarding the possible increase of motion sickness in automated vehicles. Therefore, automated vehicles without motion sickness must be developed to provide a comfortable space for drivers. In this study, we propose a control method of automated vehicles that reduces the incidence of motion sickness in passengers. In the proposed method, the desired trajectory for the lane-change task and the steering-controller improvements in a path-following controller to track the path are determined to minimize motion sickness, based on a computational model of motion sickness. Numerical simulation results demonstrate that the vehicle can be successfully controlled during the lane-change task. The effectiveness of the proposed method is demonstrated by comparing with comfort indices of previous studies.

Keywords: Human Performance Modeling, Assistive Technology, Human-Machine Interface
Abstract: Understanding and predicting the behavior of the powered wheelchair users play an important role in developing the driver assistance systems for the intelligent wheelchair. This article presents a model of wheelchair users, from the point of view of control engineering, based on the interaction between the human hand-joystick and the lumped-parameter model of the human hand muscle system. The interaction between hand and joystick is represented by a robot arm with a four-bar closed-chain mechanism. This mechanism is operated by four musculotendon units based on Hill's muscle model. This configuration allows simulating whole motions of the joystick as well as the evolutions of muscle activities during maneuvering the powered wheelchair. The advantage of this model is the integration of the biomechanical parameters of the muscular system of the human hand, which is very useful for simulating the defects related to the degree of physical impairment of wheelchair users. To validate the proposed model, we compare the simulation outputs of the user model with the previously published experimental results in the framework of the Wheelchair Skills Test (v. 4.1). The simulation results show that the proposed model can reflect the difficulties at the bio-mechanical level of users in driving the wheelchair.

Keywords: Human Performance Modeling, Human Factors
Abstract: Interest for motion sickness is increasing with increasing opportunities to view digital devices in transportation systems, including automated vehicles. Hence, technology for predicting and estimating motion sickness is essential. As one such technology, computational models for estimating motion sickness from head movements have been proposed and are used for motion sickness evaluation. However, a model capable of handling the effects of visual input and the visual-vestibular interaction for motions with six-degrees-of-freedom has yet to be developed. In this study, therefore, a computational model of motion sickness with visual-vestibular inputs is proposed by extending a model of the subjective vertical conflict theory of motion sickness for a vestibular input only. The proposed model inputs are the acceleration and angular velocity of the head and the visual perception of the angular velocity. A simulation conducted by inputting 1 h of sinusoidal oscillations demonstrated that the results are consistent with those of similar experiments conducted with human participants. In addition, the results of a simulation experiment conducted by changing the visual input demonstrated that motion sickness increases significantly when a conflict occurs between the visual and vestibular signals, which imitates the reading of a book in a moving car. Furthermore, we developed a method to calculate the predicted MSI using experimental data measured by IMU and camera image by approximating the visual perception of the angular velocity through an optical flow analysis. The results of inputting camera images and inertial measurement unit signals obtained by sine-wave-like pitching motion into the proposed model demonstrate that the proposed method, can describe the difference in motion sickness by the changes in the visual environment.

Keywords: Human Performance Modeling, Human Factors, Augmented Cognition
Abstract: Habituation responses to startle stimuli have been linked to individuals’ resilience assessment. Derived from skin conductance levels (SCL), individuals’ who elicit quicker habituation to startle stimuli typically report possessing higher trait resilience. However, this link has not yet been evaluated in the context of individuals’ dynamic resilience. This study examines whether the habituation of skin conductance response (SCR) could be used as a predictive measure for classifying individuals into high, moderate and low dynamic resilience groups. A dynamic decision-making task was paired with an acoustic startle paradigm, in which 53 participants were analysed while being subjected to acoustic startle stimuli during the task. SCL of the participants were continuously recorded during the 24-minute trials. On the contrary to previous results in the literature supporting a link between skin conductance levels and trait resilience, current results did not show a solid predictive power toward quantification of dynamic resilience. However, the results act as a first step and open the way for further research to consider testing SCL measures as predictors of dynamic resilience in more ecological task environments than in controlled laboratory settings. These more ecological task environments may yield more SCL sensitivity for dynamic resilience measurement, as individuals will experience more naturally fluctuating task load and be required to activate various combinations of cognitive functions to cope with the task stressors.

Keywords: Human Performance Modeling, Human-Machine Interface, Human-Machine Cooperation and Systems
Abstract: Sign Language Synthesis (SLS) is a domain-specific problem where multiple sign language words are stitched to generate a whole sentence in video, which serves to facilitate communications between the hearing-impaired people and healthy population. This paper presents a Variable Motion Frame Interpolation (VMFI) method for highly fluent SLS in scattered videos. Existing approaches for SLS mainly focus on mechanical virtual human technology, lacking high flexibility and natural effect. Also, the representative solutions to interpolate frames usually assume that the motion object moves at a constant speed which is not suitable for predicting the complex hand motion in frames of scattered sign language videos. To address the above issues, the proposed VMFI adopts acceleration to predict more accurate interpolated frames based on an end-to-end convolutional neural network. The framework of VMFI consists of variable optical flow estimation network and high-quality frame synthesis network that can approximate and fuse the intermediate optical flow to generate interpolated frames for synthesis. Experimental results on our realistic collected Chinese sign language dataset demonstrate that the proposed VMFI model achieves efficiency by performing better in PSNR (Peak Signal to Noise Ratio), SSIM (Structural Similarity) and MA (Motion Activity) and gets higher score in MOS (Mean Opinion Score) than other two representative methods.

Keywords: Virtual and Augmented Reality Systems, Human Performance Modeling, Human-Computer Interaction
Abstract: Hand–eye coordination (HEC) is an important ability in a variety of sports, as it involves the ability to quickly and accurately control motion using visual information. This ability is extensively involved in badminton smash reception. In existing studies on HEC evaluation, this ability was evaluated according to information collected after the motion had been performed. Accordingly, it is impossible to determine the specific phase of perception or motion that causes problems in HEC. The purpose of this study was to develop a system for identifying the abilities of perception involved in HEC as it relates to badminton smash it is considered as badminton's most effective shot. Two systems were developed to identify HEC perceptual abilities, one focused on perception in a static state (i) and the other focused on perception-to-motion comprehensiveness (ii). Next, two types of tests were conducted, one involving (i) and the other using (ii). The tests were randomly administered for a total of 18 participants in three groups: six people who had previously played badminton, six people who had played ball sports other than badminton, and six people who were inexperienced at ball sports. Participants were healthy adults. The tests aimed to identify factors of perceptual ability by comparing them across systems. The results showed no significant difference between experienced badminton players and others according to the conducted tests, where test (i) was used for measuring the distance between the shuttlecock and gaze. A significant difference between the group of experienced badminton players and the other two groups was, however, observed in the test using (ii). The results suggested the ability to follow the shuttlecock with the eyes to be a perceptual ability related to HEC in badminton smash receptions. The results of this study can help evaluate perceptual ability during smash receptions.

Keywords: Human-Machine Cooperation and Systems, Human-Machine Interface, Assistive Technology
Abstract: Dynamic positioning systems (DPSs), which maintain ships at a specific point, have been widely used in marine development; however, ships without DPSs are also required to remain at a specific position, such as during lifesaving situations. However, it is difficult to maneuver a ship to a specific point without using a maneuvering support system such as a DPS. This paper proposes a maneuvering support system that recommends the control inputs of each actuator to deck officers when they maneuver dynamic positioning manually. The support system predicts the motion of the ship and calculates the control inputs required to maintain the ship at a specific position. Subsequently, the recommended control input of each actuator is provided as a value that can be operated by deck officers. The results of the maneuvering experiment indicate that the developed system may reduce the position error in dynamic positioning.

Keywords: Human-Machine Cooperation and Systems
Abstract: The berthing maneuver requires high maneuvering techniques because the navigators have to stop the ship at a specified point with decreased maneuverability in restricted water area. Therefore, the automation of berthing maneuver has been gaining increased attention from the marine transportation industry. To construct a reliable automatic berthing system, it is desirable to apply a flexible method in regards to changing the ship’s maneuverability, similar to the human’s manual berthing maneuvering procedure. In this study, an automatic berthing maneuvering method was developed that does not require gain adjustments based on the experience of specialists or model tests, and is flexible for maneuverability changes of the ship as in manual berthing maneuvering procedure. This was accomplished by introducing an optimal control with variable gain based on the statistical model for maneuvering the ship, which was identified in real-time using actual navigation data. From the results of the experiment, it was confirmed that the method developed in this study was effective.

Keywords: Human Performance Modeling, Mental Models, Human Factors
Abstract: Numerous ship collisions occur due to navigator errors such as improper lookout in a navigational watch, which is considered an error in the situation awareness (SA) of a navigator. Moreover, mental workload is considered one of the causes of human error, and navigators are often under a high mental workload onboard, experiencing fatigue and stress. The aim of this study was to confirm that the coefficient of variation of R-R interval (CVRR) and the adaptive weighted workload (AWWL) of the National Aeronautics and Space Administration Task Load Index (NASA-TLX) are effective indexes to measure navigator fatigue. Another aim was to show the effect of fatigue on navigator SA in lookout. Thus, subjective and objective mental workload of navigators caused by fatigue was measured in bridge simulator experiments. Navigator SA was measured by using a Situation Awareness Global Assessment Technique (SAGAT). The results of the experiments were shown that CVRR was confirmed as an effective objective index for measuring navigator fatigue. Further, navigator SA errors are likely to increase under fatigue conditions.

Keywords: Human Factors, Human-Machine Interface, Kansel (sense/emotion) Engineering
Abstract: From our preceding studies, we disclosed the characteristics of the under-development device, called salivary NO₃^- Ion-Sensitive Field-Effect Transistors (NO₃^--ISFETs). These transistors have evaluated the navigators' mental workload in maneuvering a ship through the ship simulator experiments which can arrange the uniform experimental circumstances at the different simulator' s scenario. The NO₃^--ISFETs give the quantitative evaluation of the mental workload, using NO₃^- in one-drop saliva. The NO₃^--ISFETs are supposed to be used for the navigators working anywhere in a ship around the clock. Therefore, our goal is to complete the NO₃^--ISFETs, which have the durability for mucin in saliva when they measure the salivary NO₃^- long time for the purpose of the prevention of navigator’s human factors in their duties. This time, we tried to carry out another ship-simulator experiment with new NO₃^--ISFETs changed the polymer which is used at the device’s sensor part. We can confirm the instrumental difference between conventional NO₃^--ISFETs and new ones, while we evaluate the variability of navigators' mental workload.

Keywords: Team Performance and Training Systems
Abstract: It takes a great deal of time and money to turn a person with no knowledge or experience into a seafarer. However, these costs can be reduced using education and training via a simulator. In simulator-based education and training, on-site situations can be replicated, while it is also possible to freely alter both the situations of other ships and other external influences. Furthermore, this means that simulator-based training is effective not only for people with no knowledge or experience but also for those with existing experience. One of the people who supports this training is the operator, who is part of the training team. If the operator does not have a sufficient amount of skill, it is impossible for other ships to function naturally. Even if the developed training content is very good, positive training effects cannot be expected without the skill of the operator. Therefore, the operator's skills is essential to produce high-quality training for those using the technology at any level. The operators themselves can be either seafarers or non-seafarers and be placed in charge even if they do not have a license as a seafarer. However, qualifying as a seafarer does not mean that the individual will be placed in charge immediately. We think operators have a special position in the training process and represent the main key to achieving success in simulator-based training. In this study, we focused on the operators and investigated their role by comparing it with that of the captain, who is rich in knowledge and experience as a seafarer. A physiological index is designed to evaluate the mental workload, heart rate, and frequency components of the R-R interval of heart rate variability. This paper shows the mental workloads of both the navigator and the operator, while considering their mental relation using LF/HF values.

Keywords: Intelligent Power and Energy Systems
Abstract: Modern real-time systems based on heterogeneous multicore platforms can efficiently meet the disparate and high computation needs of the applications. The management of energy has become a topic of an incredible enthusiasm for analysts and practitioners during the past few years. Hence, this research presents a heuristic strategy named, CEAT, for energy-aware scheduling of a set of real-time periodic tasks on a DVFS enabled heterogeneous multicore platform. The presented strategy operates in three stages, namely Deadline Partitioning, Task-to-Core Allocation, and Energy-Aware Scheduling. Our experimental analysis shows that CEAT is not only able to successfully schedule more task sets (as high as 3.44% and 21.83%) compared to state-of-the-arts but also improve energy savings in the system.

Keywords: Intelligent Power and Energy Systems, Intelligent transportation systems
Abstract: The filter-based real-time energy management method has been proved practical and widely utilized in hybrid energy storage systems. However, the determination for the cutoff frequency of the energy-split filter is challenging. In this paper, an optimal filter-based energy management strategy is proposed for a battery/ultracapacitor electric vehicle to minimize the total energy consumption. A cost function of energy consumption for the cutoff frequency is established first. Considering the working condition of ultracapacitors, dynamic programming is adopted to obtain the optimal cutoff frequency series, i.e., the optimal energy distribution between batteries and ultracapacitors. Such an off-line optimization process is carried out under different driving cycles, e.g., urban and highway road conditions. Optimization results are used to determine the optimal cutoff frequency of a real-time filter-based energy management strategy. Simulation results indicate that the proposed strategy can minimize the total energy consumption of the hybrid energy storage system with ultracapacitors state of charge limitations being guaranteed. Compared with the existing real-time energy management strategies, the energy consumption is reduced 23.85% under aggressive acceleration conditions and 7.08% under urban conditions by the proposed strategy

Keywords: Intelligent Power and Energy Systems
Abstract: Smart grid is susceptance to a variety of cyber attacks, among which False Data Injection Attacks (FDIA) are shown to be of significantly disruptive nature. Complex, distributed, and interconnected aspects of smart grids make detection of stealthy FDIAs with high accuracy significantly challenging. To address this issue, the paper proposes an innovative Adaptive Multi-Level Features Classification for Stealthy FDIA Detection (AMLFC-SFD) based on the Alternating Current (AC) state estimation. More specifically, we focus on maintaining a trade-off between the accuracy rate of detection and its associated computational complexity by utilizing two different Support Vector Machine (SVM)-based classifiers, in which the number of features as the input of the classifier depends on the strength of the underlying attack. In this regard, we divide potential FDI attacks in smart grids into three decision regions, including strong, moderate, and weak attacks and obtain the most accurate Kernel to separate measurements. To evaluate the proposed AMLFC-SFD framework, comprehensive numerical experiments are performed based on the IEEE 30-bus system. Results illustrate that with a lower number of features a reasonably high detection accuracy can be achieved, leading to a considerably less run time, which is of paramount importance for practical implementation.

Keywords: Intelligent Power and Energy Systems, Distributed Intelligent Systems, Smart urban Environments
Abstract: This paper proposes a novel robust decentralized charging strategy for large-scale EV fleets. The system incorporates multiple EVs as well as inelastic loads connected to the power grid under power flow limits. We aim at minimizing both the overall charging energy payment and the aggregated battery degradation cost of EVs while preserving the robustness of the solution against uncertainties in the price of the electricity purchased from the power grid and the demand of inelastic loads. The proposed approach relies on the so-called uncertainty set-based robust optimization. The resulting charge scheduling problem is formulated as a tractable quadratic programming problem where all the EVs’ decisions are coupled via the grid resource-sharing constraints and the robust counterpart supporting constraints. We adopt an extended Jacobi-Proximal Alternating Direction Method of Multipliers algorithm to solve effectively the formulated scheduling problem in a decentralized fashion, thus allowing the method applicability to large scale fleets. Simulations of a realistic case study show that the proposed approach not only reduces the costs of the EV fleet, but also maintains the robustness of the solution against perturbations in different uncertain parameters, which is beneficial for both EVs' users and the power grid.

Keywords: Intelligent Power and Energy Systems, Decision Support Systems
Abstract: Most Distribution Systems (DS) are set to operate in a radial configuration for many reasons, such as economy, security etc. The Distribution Network Reconfiguration (DNR) is the cheapest and commonly way used to obtain better topology to achieve many different objectives. This paper presents an efficient methodology based on Graph Theory for imposing the radial topology constraint in unfeasible solutions that appear in Evolutionary Algorithms during the evolution of the optimal solution. To reach a better computational time, this paper also presents an effective Backward-Forward method to solve power flow in radial DS (RDS). The algorithm developed was tested in well-known distribution systems.

Keywords: Space Systems, Robotic Systems, Intelligent Learning in Control Systems
Abstract: This paper develops a constrained Model Predictive Control (MPC) formulation for longitudinal control of a fixed wing aircraft. In order to account for the inherent coupling between the inputs and outputs, which is a characteristic of multivariable systems, multiple prediction horizons are used, i.e. one for each output. Furthermore, a novel Quadratic Programming (QP) problem is derived to solve this MPC problem, via the Primal-Dual procedure. Numerical simulations using two QP algorithms demonstrate successful tracking performance of the MPC based controller.

Keywords: Intelligent Learning in Control Systems, Cooperative Systems, Robotic Systems
Abstract: Robots are increasingly exploited in production plants, with the need to learn and to adapt themselves to new tasks. This paper focuses on the investigation of machine learning techniques to make a sensorless robot able to learn and optimize an industrial assembly task. Relying on sensorless Cartesian impedance control, a task-trajectory learning algorithm exploiting a limited number of human’s demonstrations (based on Hidden Markov Model), and an autonomous optimization procedure (based on Bayesian Optimization) are proposed to learn and optimize the assembly task. To validate the proposed methodology, an assembly task of a gear into its square-section shaft has been considered. A Franka EMIKA Panda manipulator has been used as a test platform. The experiments show the effectiveness of the proposed strategy, making the robot able to learn and optimize its behaviour to accomplish the assembly task, even in the presence of uncertainties.

Keywords: Decision Support Systems, Intelligent Learning in Control Systems, Robotic Systems
Abstract: Reinforcement Learning (RL) platforms play an important role in translating the rapid development of RL algorithms into the successes of real-world tasks. These platforms integrate multiple simulation environments, allowing testing, evaluating and finally applying RL algorithms in different scenarios. However, the algorithm code is required to execute in the same runtime system with the underlying environments, which limits platforms' compatibility when adapting an algorithm and flexibility when switching between different algorithms. We propose Glue to resolve this issue, by decoupling the executions of algorithms and environments first, then leveraging the RPC protocol to orchestrate a seamless workflow between them. Glue is further implemented as a library, which hides the handling of language-specific RPCs from users. We evaluate Glue by adapting 6 RL algorithm implementations to a representative RL platform. Compared with the baseline approach, Glue enables algorithms to achieve competitive performance, but reduces lines of algorithm code to be changed in adaption by 27.77% , at the cost of 5.40% longer training time, on average.

Keywords: Intelligent Learning in Control Systems, Robotic Systems
Abstract: Underactuated systems like sea vessels have degrees of motion that are insufficiently matched by a set of independent actuation forces. In addition, their trajectory-tracking control problem grows in complexity in order to decide the optimal rudder and thrust control signals. This enforces several difficult-to-solve constraints that are associated with the error dynamical equations using the classical optimal tracking and adaptive control methods. An online machine learning mechanism based on integral reinforcement learning is proposed to find a solution for a class of nonlinear tracking problems with partial prior knowledge of the system dynamics. The actuation forces are decided using innovative forms of temporal difference equations relevant to the vessel's surge and angular velocities. The solution is implemented using an online value iteration process which is realized by employing means of the adaptive critics and gradient descent approaches. The adaptive learning mechanism exhibited well-functioning and interactive features in react to different desired reference-tracking scenarios.

Keywords: Robotic Systems
Abstract: This paper presents a structural recognition system using machine learning algorithms (Multilayer Perceptron, Support Vector Machine and Random Forest) and the environment information to analyzes the feasibility of the use of machine learning methods for the construction of topological maps. The proposed method combines the recognized information from a given scene with a topological graph to create a map. This map can be used to plan high-level tasks of robotic navigation. The topological nodes are used to store semantic information, such as the robot’s poses, sensor data and scene characteristics. The machine learning algorithms classification of the structural information as either rooms, corridors or doors obtained a satisfactory performance. The topological maps built efficiently from structural recognition provided by classification

Keywords: Robotic Systems
Abstract: Robotic path planning must avoid obstacles or singularities, and to check for such constraints periodically is computation and time intensive. This paper introduces Constraint-free Discretized Manifolds for robotic Path planning (CDMP) to formulate a constraint free space in the configuration and work spaces. Application based necessities such as obstacle and path visualization, or computational ease guides the choice of working in either space. The chosen constraint free manifold is then meshed using DistMesh, for use with path planning algorithms, which in this paper is A*. The merits offered by this solution are two fold- first, the formulated constraint free manifold is guaranteed to be singularity free irrespective of the start and end locations; second, the path chosen on this manifold will be the shortest path by virtue of using A*.

Keywords: Robotic Systems, Technology Assessment
Abstract: With the emergence of agile manufacturing in highly automated industrial environments, the demand for efficient robot adaptation to dynamic task requirements is increasing. For assembly tasks in particular, classic robot programming methods tend to be rather time intensive. Thus, effectively responding to rapid production changes requires faster and more intuitive robot teaching approaches. This work focuses on combining programming by demonstration with path optimization and local replanning methods to allow for fast and intuitive programming of assembly tasks that requires minimal user expertise. Two demonstration approaches have been developed and integrated in the framework, one that relies on human to robot motion mapping (teleoperation based approach) and a kinesthetic teaching method. The two approaches have been compared with the classic, pendant based teaching. The framework optimizes the demonstrated robot trajectories with respect to the detected obstacle space and the provided task specifications and goals. The framework has also been designed to employ a local replanning scheme that adjusts the optimized robot path based on online feedback from the camera-based perception system, ensuring collision-free navigation and the execution of critical assembly motions. The efficiency of the methods has been validated through a series of experiments involving the execution of assembly tasks. Extensive comparisons of the different demonstration methods have been performed and the approaches have been evaluated in terms of teaching time, ease of use, and path length.

Keywords: Robotic Systems
Abstract: This paper presents a multi-objective vibration-based particle-swarm-optimization (MO-VBPSO) algorithm with enhanced exploration ability and convergence performance, for training fuzzy-controller (FC) to achieve robot control. The MO-VBPSO applies a reference point-based leader selection schema that assigns leaders for MO-PSOs’ searching. Besides, the MO-VBPSO framework is integrated with a vibration factor to strengthen the exploration ability for resolving the local minima issue, which is inspired by the amplitude of the Firework Algorithm (FWA). The evaluation of MO-VBPSO focuses on the effect of the vibration factor by applying it to training a mobile robot in a simulation environment. The evaluation results are discussed concerning exploration ability, convergence performance, and performance stability. Experimental results reveal that the proposed MO-VBPSO lifts the performance of robot training significantly.

Keywords: Robotic Systems
Abstract: Abstract—Traversing at a high speed while maintaining stability is important for the application of quadruped robots. Prior works mainly concentrated on optimizing the stability margin of quadruped robots when walking through a variety of terrains. However, the problem of improving quadruped robots’ velocity with the static gait is less concerned in their works. In this paper, the static gait planning problem is considered under the assumption that a set of irregular footholds on the rough terrain are given, and two approaches are proposed to improve the walking speed. The first one is a distance optimization algorithm, which can minimize the moving distance of the center of gravity (COG) in the stance phases based on the stability and the kinematic constraint. The other is a velocity optimization algorithm, which enables the body and the feet to move at the highest velocity with the joint angular velocity limit. The joint application of these two optimizations significantly improves the static walking speed of the quadruped robot. Simulation results in V-REP are presented to demonstrate the effectiveness of the proposed approaches in improving the walking speed. Compared with the traditional gait planning techniques, one that moves the robot with the optimal stability margin, and the counterpart of the proposed method without optimizing the velocity, our algorithms increase the average walking velocity by 94.2% and 32.7%, respectively.

Keywords: Model-based Systems Engineering, Robotic Systems
Abstract: Robots are currently mostly found in industrial settings. In the future, a wider range of environments will benefit from their inclusion. This calls for the development of tools that allow professionals to set up dependable robotic applications in which people productively interact with robots aware of their needs. Given the co-existence of humans and robots, the precise analysis—e.g., through formal verification techniques—of properties related to aspects such as human needs and physiology is of paramount importance. In this paper, we present a formally-based, model-driven approach to design and verify scenarios involving human-robot interactions. Some of the features of our approach are tailored to the healthcare domain, from which our case studies are derived. In our approach, the designer specifies the main parameters of the mission to generate the model of the application, which includes mobile robots, the humans to be served, including some of their physiological features, and the decision-maker that orchestrates the execution. All components are modeled through hybrid automata to capture variables with complex dynamics. The model is verified through Statistical Model Checking (SMC), using the Uppaal tool, to determine the probability of success of the mission. The results are examined by the developer, who iteratively refines the design until the probability of success is satisfactory.

Keywords: Robotic Systems, Decision Support Systems, Conflict Resolution
Abstract: Behavior Trees (BTs) constitute a widespread artificial intelligence tool that has been successfully adopted in robotics. Their advantages include simplicity, modularity, and reusability of code. However, Behavior Trees remain a high-level decision making engine; control features cannot easily be integrated. This paper proposes Reconfigurable Behavior Trees (RBTs), an extension of the traditional BTs that incorporates sensed information coming from the robotic environment in the decision making process. We endow RBTs with continuous sensory data that permits the online monitoring of the task execution. The resulting stimulus-driven architecture is capable of dynamically handling changes in the executive context while keeping the execution time low. The proposed framework is evaluated on a set of robotic experiments. The results show that RBTs are a promising approach for robotic task representation, monitoring, and execution.

Keywords: Robotic Systems, Model-based Systems Engineering, System of Systems
Abstract: Companies producing Robot-Like Systems (RLS) must increase efficiency in design and production in order to stay competitive in the international market. Such RLS range from small SCARA robots to entire production facilities. Very different systems for similar tasks are published in research or offered in the market. A clear path to an optimal configuration for a given task is apparently not available to the engineer. This indicates that the gear drives of such RLS and their integration with automation technology are still insufficiently researched, and their efficient design poses significant challenges for the industry. This paper identifies and provides a concise overview of requirements and challenges in the context of drives for such RLS. Building on this overview, suggestions are made for the future course of action in research.

Keywords: Robotic Systems
Abstract: The continuum robot has attracted more attention for its flexibility. Continuum robot kinematics models are the basis for further perception, planning, and control. The design and research of continuum robots are usually based on the assumption of piecewise constant curvature (PCC). However, due to the influence of friction, etc., the actual motion of the continuum robot is approximate piecewise constant curvature (APCC). To address this, we present a kinematic equivalent model for continuum robots, i.e. APCC 2L-5R. Using classical rigid linkages to replace the original model in kinematic, the APCC 2L-5R model effectively reduces complexity and improves numerical stability. Furthermore, based on the model, the configuration self-estimation of the continuum robot is realized by monocular cameras installed at the end of each approximate constant curvature segment. The potential of APCC 2L-5R in perception, planning, and control of continuum robots remains to be explored.

Keywords: Robotic Systems
Abstract: Coordination and control of actions using sensors on the robot is the key technology for robots to become intelligent. Aiming at the communication and control problems of existing sensor robots, an object-oriented data communication framework based on OIO middleware is proposed. Through the research of serialization and deserialization of communication data, the problems of coordinated advancement, climbing, and turning of sensor robot clusters are explored, and the law of object data and intelligent control instruction transmission is revealed. The research in the paper provides new ideas for the development of sensor robot motion coordination and control modules, and promotes the intelligentization of sensor robots. The paper designs OIO middleware that can be used for sensor robot cluster communication and control, and regards the sensor robot as an object that encapsulates attributes and methods, and performs intelligent control based on object-oriented data. Experimental research on dual-robot and multi-robot collaborative climbing, collaborative turning, and collaborative obstacle avoidance based on OIO middleware was carried out. The experimental results show that this method simplifies the communication and control of the robot cluster and improves the control effect.

Keywords: Robotic Systems
Abstract: Reinforcement learning algorithms have been successfully trained for games like GO, Atari, and Chess in simulated environments. However, in cue sport-based games like Carrom, real world is unpredictable unlike in Chess and GO due to the stochastic nature of the gameplay as well as the effect of external factors such as friction combined with multiple collisions. Hence, solely training in a simulated platform for games like Billiard and Carrom, which need precise execution of a shot, would not be ideal in actual gameplay. This paper presents a real-time vision based efficient robotic system to play Carrom against a proficient human opponent. We demonstrate the challenges of adopting a reinforcement learning algorithm beyond simulations in implementing a strategic gameplay for the robotic system. We currently achieve an overall shot accuracy of 70.6% by combining heuristic and reinforcement learning algorithms. Analysis of the overall results suggests the possibility of adopting a real-world training for board games which need precise mechanical actuation beyond simulations.

Keywords: Intelligent Learning in Control Systems, Robotic Systems, Grey Systems
Abstract: Reinforcement learning (RL) is an effective method to design robust control. Uncertainty in the worst case requires large state-action learning space. The continuous time RL can solve this computational problem. In this paper, we modify the classical continuous time RL. Compared with the actor-critic (AC) algorithm, our method is more simple and more robust under the worst-case uncertainty.

Keywords: Robotic Systems, Intelligent Learning in Control Systems
Abstract: Following its applications on reception, space exploration, health care, and others, the dual-arm robot is employed for industrial automation more intensively these days. With a human-like two-arm structure of high degrees of freedom, it should be a strong competitor for industrial tasks of high complexity. However, up to date, its deployment is still far behind the traditional single-arm robot manipulator. It may be partially due to its higher price and larger installation space. Meanwhile, the lack of proper methods or devices to deal with task planning and teaching that usually involves robot motions of more than 10 DOF is also influential. Motivated by it, in this paper, we propose a novel manipulation system for the industrial dual-arm robot manipulator using virtual reality (VR). Especially, we propose taking advantage of the similarity between the dual-arm robot and human arm and come up with a human-like path planner, with an intention to incorporate that of human into robot path planning. In addition, we also furnish the system with several assistive tools, including that for physical behavior emulation and self-collision detection. Simulations based on realistic environments are conducted to demonstrate the effectiveness of the proposed manipulation system, along with the questionnaires to evaluate user’s responses.

Keywords: Intelligent Learning in Control Systems
Abstract: This research proposes a double function-link fuzzy brain emotional controller (DFLFBC) to synchronize a 4D hyper-chaotic system. The DFLFBC contains three main structures and a set of fuzzy inference rules. Three main structures comprises a double function-link network, an amygdala network, and an orbitofrontal cortex network. The double function-link is employed for adjusting the output weights of the amygdala and orbitofrontal cortex networks that imitate the functions of the amygdala and orbitofrontal cortex of a brain. The online learning laws of the parameters of the system are derived from the gradient descent algorithm. Synchronization studies of a 4D hyper-chaotic system are implemented to illustrate the capability and performance of the proposed DFLFBC.

Keywords: Intelligent transportation systems, Intelligent Assistants and Advisory Systems
Abstract: he visual perception is of great significance for advanced driving assistance systems or autonomous driving vehicles to recognize the surrounding scenes. In the adaptation to the real environments for collision warnings, a sensor system should be efficient and has the strong ability to detect small objects. This paper presents a forward collision warning technique which incorporates the object detection and depth estimation networks. A deep convolutional neural network is constructed with transfer connection blocks for object detection and classification. It is capable of small object detection under the real-time processing requirement. For depth estimation, a monocular based disparity estimation network is adopted to the stereo vision framework. The epipolar constraint is applied to increase the prediction accuracy. In the experiments, the performance evaluation is carried out on public driving datasets. The comparison with the state-of-the-art networks has demonstrated the feasibility of the proposed technique.

Keywords: Cyber-Physical Cloud Systems, Service Systems and Organization
Abstract: Container-based virtualization is becoming a preferred choice to deploy services since it is lightweight and supports on-demand scalability as well as availability. The Process Automation Industry has accepted this technology to make their applications service oriented. However, container-based microservice architecture is effective only when the original software strictly followed modularity principles during its design. In this article, we share our learning of converting a distributed software to a microservice-based architecture using containers. Though the existing system has a modular design and deployed as distributed components, analysis of the current architecture shows that the application is monolithic (though modularized) and the components are strongly coupled in an indirect manner. As a result, it turns to be impossible to attain microservice-based architecture without changing the architecture. Next, we propose a microservice-based containerized TO-BE architecture of the application, and demonstrate that this TO-BE architecture does not incur any significant overhead. Finally, we propose a set of recommendations that the practitioners can follow to convert a monolithic application to a containerized architecture.

Keywords: Distributed Intelligent Systems, Cooperative Systems, Intelligent Green Production Systems
Abstract: Until quite recently, multi-agent systems (MAS) and corresponding technology have been an area of high expectations of industrial IT community. However, in the reality, these expectations are still not met and, in practice, the industry very rarely uses the MAS design methodologies, technologies and software tools despite their existence in many variants and, what can seem more surprisingly, despite appearance of many new classes of applications for which the MAS paradigm could be the perfect match. The paper analyzes the root causes of the mismatch between the recent industrial anticipations and real state of the practical use of MAS. It identifies engineering problems with very little re-use of code that currently stops economics of scale and impede the extensive industrial deployment of MAS and the ways to overcome them.

Keywords: Computational Intelligence, Expert and Knowledge-based Systems, Machine Learning
Abstract: Unstructured data represent as much as 90% of all business-relevant information. In Brazil, the practice of printing official journals dates back to the 19th century. Today more than 200 official journals in circulation, which together accumulate around 1.4 billion publications without textual standard. This work proposes the development of a system for unstructured data mining using a Dynamic Ensemble Selection. JudEasy implements, added in addition to classic text pre-processing methods, a set of twelve DES and a static method for creating categorized textual models for Brazilian of official journals. As results the DES-KL model obtained the highest accuracy rate of 96.51% and exceptional precision of 0.99

Keywords: Information Systems for Design/Marketing, Web Intelligence and Interaction
Abstract: The number and variety of Web APIs is growing exponentially. Software engineers need to expend a significant amount of time and effort reading and understanding the accompanying documentation. In addition, system automation may use API to interact with each other. However, this is not always a simple task since the API documentation of a provider can be anything from a single HTML page description through to a complex structure with information spanning several pages. Understanding this wide variety of API documentation structures and styles is therefore a labor intensive and error-prone task for engineers. By providing a machine-learning platform that can extract and standardize API usage information, however, we believe we can accelerate the creation of API-enabled systems by using automation to simplify the task of understanding. In this paper, we introduce a novel approach to automating and standardizing usage information about APIs, combining several machine-learning algorithms in order to extract key attributes from API documentation and generate a machine readable Open API Specification (OAS). We develop i) a content-based learning model that identifies the context of a block of extracted API features; ii) a signature-based machine-learning model that recognizes a sequence of successful/unsuccessful extracted API endpoints; and iii) a deep mapping model that pinpoints fine-grained mapping of extracted API attributes to OAS objects. Results of our experiments show that the proposed approach successfully works with an accuracy of 99%, 94% and 97%for content-based learning, signature-based learning, and Deep Mapping of API attributes respectively. We then use the models to produce OAS compliant API Specifications for more than 2,585 public APIs, validate them via API calls and finally deploy the validated APIs to the RunMyProcess software automation platform.

Keywords: Knowledge Acquisition in Intelligent, Optimization, Machine Learning
Abstract: The task of predicting energy consumption is a problem of great interest within the context of electric power companies. A minimal error prediction is important for identifying inconsistencies in the monthly consumption reading process. This paper presents a methodology applied to the prediction of electric consumption. This was performed with and without a hyperparameter optimization strategy using a TCN network. These strategies were applied to individual electric consumption time series. The TCN approach had superior results when compared to SES, ARIMA, and Gradient Boosting. The results show that the proposed process obtained low efficiency with approximately 1% or less improvement compared to the use of no optimization but the TCN itself showed promising results being the best approach in many of ours tests.

Keywords: Human-Machine Interface, Brain-based Information Communications, Assistive Technology
Abstract: Recently, the complexity analysis of brain activity has shown the possibility to provide additional information to discriminate between rest and motion in real-time. In this work, we propose a novel entropy-based machine learning method to classify between standing and walking conditions from the sole brain activity. The Shannon entropy has been used as a complexity measure of electroencephalography (EEG) signals and subject-specific features for classification have been selected by Common Spatial Patterns (CSP) filter. Exploiting these features with a linear classifier, we achieved >85% of classification accuracy over a long period (~25 min) of standing and treadmill walking on 11 healthy subjects. Moreover, we implemented the proposed approach to successfully discriminate in real-time between standing and over-ground walking on one healthy subject. We suggest that the reliable discrimination of rest against walking conditions achieved by the proposed method may be exploited to have more stable control of devices to restore locomotion, avoiding unpredictable and dangerous behaviors due to the delivery of undesired control commands.

Keywords: Human-Machine Interface, Assistive Technology
Abstract: Stroke is a neurological syndrome that may affect upper and lower limbs functions of post-stroke survivors. Brain-Computer Interfaces (BCIs) are becoming as a promising alternative to help post-stroke patients rehabilitation, although there are very few associated studies and systems being applied in clinical environment. As a novelty, developing a motor imagery (MI) BCI based on pedal end-effector for motor rehabilitation, we propose to combine pedaling MI and passive pedaling into a Calibration phase. As a result, users would activate continuously their central and peripheral mechanisms linked to lower-limbs throughout BCI intervention. We hypothesize that this strategy enables to obtain a better classification model for our BCI by selecting those feature vectors corresponding to pedaling MI closer to real movements. Therefore, it is expected to have a more effective BCI intervention. Preliminary results show that the proposed method may increase the BCI performance. For almost all participants was noted, during MI tasks, a power decreasing over the foot area (Cz location), corresponding mainly to beta frequency bands, specifically for both low (13 to 22 Hz) and high (23 to 30 Hz) beta bands.

Keywords: Brain-based Information Communications, Human-Computer Interaction, Human-Machine Interface
Abstract: Typical EEG-based BCI applications require the computation of complex functions over the noisy EEG channels to be carried out in an efficient way. Deep learning algorithms are capable of learning flexible nonlinear functions directly from data, and their constant processing latency is perfect for their deployment into online BCI systems. However, it is crucial for the jitter of the processing system to be as low as possible, in order to avoid unpredictable behaviour that can ruin the system's overall usability. In this paper, we present a novel encoding method, based on on deep convolutional autoencoders, that is able to perform efficient compression of the raw EEG inputs. We deploy our model in a ROS-Neuro node, thus making it suitable for the integration in ROS-based BCI and robotic systems in real world scenarios. The experimental results show that our system is capable to generate meaningful compressed encoding preserving to original information contained in the raw input. They also show that the ROS-Neuro node is able to produce such encodings at a steady rate, with minimal jitter. We believe that our system can represent an important step towards the development of an effective BCI processing pipeline fully standardized in ROS-Neuro framework.

Keywords: Human-Machine Interface, Assistive Technology, Brain-based Information Communications
Abstract: Brain-machine interfaces (BMI) open new opportunities to control robotic devices as they provide the feasibility to translate brain signals into commands. Severely motor impaired people who have lost muscle control could benefit from this technique to control assistive devices, which support them in daily life. However, non-invasive BMIs can distinguish only a few different commands with relatively high error rates, which makes the asynchronous control of a robot with multiple degrees of freedom challenging. Here, we introduce a novel robotic grasping system, which combines scene recognition techniques and autonomous path planning with user interaction instantiated by a hybrid control system based on the electroencephalogram and the electrooculogram. The results show that healthy subjects can reliably perform a grasp-and-place task, arranging four objects at defined positions within 133-331s (193.6 ±61.5s), while they require only a few corrections. Our robot control platform proved to work solely with electrophysiological control signals and thus, constitutes a basis to perform various robot actions initiated by motor-impaired people.

Keywords: Human-Machine Interface, Human-Computer Interaction, Assistive Technology
Abstract: The increasing interest of the research community in the intertwined fields of brain-machine interface (BMI) and robotics has led to the development of a variety of brain-actuated devices, ranging from powered wheelchairs and telepresence robots to wearable exoskeletons. Nevertheless, in most cases, the interaction between the two systems is still rudimentary, allowing only an unidirectional simple communication from the BMI to the robot that acts as a mere passive end-effector. This limitation could be due to the lack of a common research framework, facilitating the integration of these two technologies. In this scenario, we proposed ROS-Neuro to overcome the aforementioned limitations by providing a common middleware between BMI and robotics. In this work, we present a working example of the potentialities of ROS-Neuro by describing a full closed-loop implementation of a BMI based on motor imagination. The paper shows the general structure of a closed-loop BMI in ROS-Neuro and describes the specific implementation of the packages related to the proposed motor imagery BMI, already available online with source codes, tutorials and documentations. Furthermore, we show two practical case scenarios where the implemented BMI is used to control a computer game or a telepresence robot with ROS-Neuro. We evaluated the performance of ROS-Neuro by ensuring comparable results with respect to a previous BMI software already validated. Results demonstrated the correct behavior of the provided packages.

Keywords: Evolutionary Computation, Optimization, Machine Learning
Abstract: With the advance in technology, the volume of available data grows massively. Therefore, feature selection has become an essential preprocessing step to extract valuable information. Feature selection is the task of reducing the number of features by removing redundant features from data while preserving the classification accuracy. It is a multiobjective problem as there are two objectives. In general, multiobjective selection algorithms with machine learning techniques are utilized to find the most promising feature subsets; however, classification performances of these machine learning techniques are analyzed separately. In this study, we propose a new multiobjective selection model that dynamically searches for the best ensemble of five classifiers to extract the best representative feature subsets. We present the experiment results on 12 well-known datasets. The results show that the proposed method performs significantly better than all the machine learning techniques when they are executed separately. Moreover, the proposed method outperforms two existing ensemble algorithms, namely AdaBoost and Gradient Boosting.

Keywords: Evolutionary Computation, Optimization, Swarm Intelligence
Abstract: Multimodal optimization problems (MMOPs) require the algorithm to locate multiple global optima and also achieve a certain accuracy on the found optima. When applying particle swarm optimization (PSO) to solve MMOPs, a fixed population communication topology may not be sufficient to handle these two requirements simultaneously. In this paper, a novel PSO with hybrid ring topology, termed HRTPSO, is proposed for MMOPs. In the early evolutionary process of HRTPSO, a sparse topology is constructed to enhance the population diversity to help locate multiple optima, while in the later evolutionary process of HRTPSO, the population communication topology is switched to a relatively dense topology for improving the convergence efficiency on the found optima. The switch of topology is controlled by a threshold and its effect is also analyzed in this paper. Experimental results on the 20 multimodal functions in CEC’2013 benchmark set show that HRTPSO has better performance than the other six multimodal optimization algorithms.

Keywords: Evolutionary Computation, Swarm Intelligence, Computational Intelligence
Abstract: Particle Swarm Optimization (PSO) is a swarm intelligence method which is employed frequently for solving real-world problems. After its inception, many variants of PSO devote to improving its performance by modifying the behavior of each particle, in which the population topologies of the particle swarm may alter. This paper investigates how population topology influences the performance of PSO. A random topology generation algorithm that adopts both the greedy strategy and randomized algorithm is proposed in the paper. The randomly generated topologies are applied in PSO-w, which introduces no modification to the population topology of the original PSO. Experimental results demonstrate that algorithms using topologies with more edges tend to converge faster and generally obtain a more accurate solution. Another major result in this paper is that how clustering coefficient affects PSO largely depends on the sparsity of the topology. A lower clustering coefficient in sparse topology conduces to faster convergence and a more precise result, but a higher clustering coefficient is preferred when the topology is dense.

Keywords: Evolutionary Computation, Swarm Intelligence, Optimization
Abstract: The allocation of epidemic-control resources has been an increasingly active topic in the physical world. Most existing studies focus on the allocation of abstract and continuous epidemic control resources, and then formulate differentiable convex programming problems. However, real-world resources are usually discrete materials, goods, or services, so that resource allocation problems become non-convex. As a complementary study, this paper builds three discrete resource allocation problems based on an improved Susceptible-Exposed-Infectious-Vigilant (SEIV) spread model: the cost-constraint optimization problem (CCOP), rate-constraint optimization problem (RCOP), and eradication optimization problem (EOP). Then, existing swarm-based metaheuristic algorithms are adapted to effectively solve the problems. Thereinto, the Heuristic Majority-Voting Binary Particle Swarm Optimizer (HMV-BPSO) is present, which introduces a heuristic factor which concerns the probability distribution of resources to guide the evolution of particles and helps improve the performance of original MV-BPSO. Numerical experiments are developed to verify the effectiveness of swarm-based metaheuristic algorithms on epidemic control. Results show that HMV-BPSO can produce higher-quality solutions than other algorithms.

Keywords: Evolutionary Computation, Swarm Intelligence, Optimization
Abstract: Particle swarm optimization (PSO) has demonstrated a promising performance for solving challenging optimization problems, but its performance in solving large-scale optimization problems (LSGO) has drastically decreased. In the canonical PSO, velocity has a significant effect on the performance of PSO, which is updated based on cognitive and social factors. It can help particles to share information effectively. In this paper, a center-based velocity is proposed in which a new component, named opening ”center of gravity factor”, is added to velocity update rule to propose the center-based PSO (CenPSO). Center of gravity factor benefits from center-based sampling strategy, a new direction in population-based metaheuristics, especially to tackle LSGOs. The proposed method is evaluated on two benchmark functions, namely, CEC2010 and CEC2017, with dimensions 100 and 1000. The experimental results verify that CenPSO is significantly better than PSO over the majority of benchmark functions.

Keywords: Image Processing/Pattern Recognition
Abstract: With the widespread use of image processing and computer vision applications, effective denoising methods are highly sought after, prompting the development of a variety of algorithms under different assumptions on noise and signal properties. However, most of these techniques are developed to deal with grayscale images, and are typically extended to colour images by processing each RGB channel separately. In this paper, we extend the curvelet power shrinkage algorithm, introduced previously for grayscale images, to colour image denoising, by applying the proposed method in the luminance/opponent-colour YCbCr colour space to take into consideration image inter-channel dependencies. The performance of the proposed algorithm on colour images corrupted by additive white Gaussian noise is evaluated in terms of both objective and subjective measures, and the obtained results show our method to be competitive to other methods including curvelet domain hard thresholding and MSt-SVD.

Keywords: Image Processing/Pattern Recognition
Abstract: The human visual system receives sensory inputs from the environment and converts them into the perception of real objects such as desks, buildings, and cars. This phenomenon is known as visual perception. It is an effortless process which makes sense of the visual information. Recently, the ultimate goal for machine vision researches is to imitate the human visual perception and to understand the intricate data processing done by the human brain. In this paper, we propose a methodology inspired by Gestalt theory of perception to quantify the human visual perception. Our goal is to add a quantitative aspect for the visual perception in order to be easily integrated into the image processing tasks. Tests on synthetic images show good performance on images with Gaussian noise proving its efficiency to detect perceptual groups.

Keywords: Image Processing/Pattern Recognition
Abstract: Clustering is one of the most important methods to discover the intrinsic grouping in a set of unlabeled data. As ways of getting data are more various and easier, the amount of data processed is increasing exponentially and the data is more likely to be located at different clients. Traditional clustering methods cannot process the large dataset one time due to the limit of memories. In this paper, an Image Scaling Density-based Clustering (ISDC) algorithm is proposed. ISDC can process data by a client alone as well as process in parallel by several clients to deal with data located at different clients. The ISDC algorithm does not need any parameters to be designated manually. The parameters are determined by the algorithm based on the statistical features of dataset. In Parallel ISDC or PISDC, each data block located at different client is clustered alone to form intermediate clusters. By border detection algorithm, representative clusters are formed by the points that are at the edge of intermediate clusters. Then, in global clustering, representative clusters from all clients are merged by the server. The border detection algorithm reduces the communication cost between clients and the server, as well as increases the efficiency of global clustering. At last, the server feeds back the clustering information to clients to complete clustering. Our experimental results verified the effectiveness and efficiency of PISDC and ISDC.

Keywords: Image Processing/Pattern Recognition, Fuzzy Systems and their applications, Machine Learning
Abstract: In this paper we introduce the DMR -- a prototype-based method and network architecture for deep learning which is using a decision tree (DT)- based inference and synthetic data to balance the classes. It builds upon the recently introduced xDNN method addressing more complex multi-class problems, specifically when classes are highly imbalanced. DMR moves away from a direct decision based on all classes towards a layered DT of pair-wise class comparisons. In addition, it forces the prototypes to be balanced between classes regardless of possible class imbalances of the training data. It has two novel mechanisms, namely i) using a DT to determine the winning class label, and ii) balancing the classes by synthesizing data around the prototypes determined from the available training data. As a result, we improved significantly the performance of the resulting fully explainable DNN as evidenced on the well know benchmark problem Caltech-101. Furthermore, we also achieved high results in terms of accuracy for the well known Caltech-256 dataset, as well as surpassed the results of other approaches on Faces-1999 problem. In summary, we propose a new approach specifically advantageous for imbalanced multi-class problems on well known hard benchmark datasets. Moreover, DMR offers full explainability, does not require GPUs and can continue to learn from new data by adding new prototypes preserving the previous ones but not requiring full retraining.

Keywords: Image Processing/Pattern Recognition, Industry 4.0
Abstract: Action extraction from teleoperated robots can be a crucial step in the direction of full -or shared- autonomy of tasks where human experience is indispensable. This is especially important in tasks that seek dynamic goals, where a human operator needs more control on how the machine behaves to provide assistance or perform a task. Open-die forging is a basic metal-forming process that lacks non-destructive product quality measures. Human experience is therefore imperative. During the process, a robot-arm is operated to place the work-piece between the dies of the forge where it is striked several times to reach a specific geometry. In this paper, we apply a white-box computer vision technique to discretize open-die forging robot-arm teleoperation data into actions as a step in learning the operator's behavior.

Keywords: Image Processing/Pattern Recognition, Machine Learning, Industry 4.0
Abstract: Vegetable mass estimation from monocular RGB camera images is proposed. Vegetables are fragmented and placed on a conveyor belt of food processing machine and the monocular camera placed over the belt take pictures of vegetables on the belt. The proposed system does not employ any scale, load cell, and other mass scaling equipment. We apply pre-trained convolutional neural networks to estimate the mass of vegetables. Transfer learning including various levels of fine-tuning is also applied. For pre-trained network, we use Xception, VGG16, ResNet50, and Inception_v3, which are pre-trained using ImageNet. The result shows that the best estimation accuracy is achieved by VGG16, whose MAPE (mean average percentage error) is 11.1%. Additionally, we fine-tune VGG16 and the accuracy reduces to 7.9% for MAPE. From this result, the performance of CNN model can improve by fine-tuning. The proposed system can be applied to low-cost, high-speed, and efficient measurement of foods replaced to load cells.

Keywords: Industry 4.0
Abstract: The increasing influence and adoption of Additive Manufacturing (AM) technology across manufacturing sectors has made it a target for cyber-physical attacks. While several techniques have been developed to mitigate specific AM related threats, there is little research aimed at assessing cyber-physical threats or vulnerabilities that extend across entire AM workflow. Such an assessment requires a holistic approach that systematically analyzes all components of the AM workflow for cyber-physical vulnerabilities. Several methodologies have been successfully applied towards identifying such vulnerabilities in other domains like Information Technology (IT) systems. In response, this paper provides a systems approach towards identifying cyber-physical vulnerabilities in AM systems using the Vulnerability Assessment and Mitigation (VAM) methodology. This approach characterizes the different vulnerabilities that arise from specific AM threats by identifying the quality attributes of the AM system that introduces it. The security techniques developed to mitigate these threats are reduced to a combination of fundamental mitigation techniques, that have been compiled based on its success in other domains. Using the resources from the VAM methodology, fundamental mitigation techniques that are best suited towards mitigating specific vulnerability attributes are identified. Comparing the combination of fundamental mitigation techniques associated with a AM security technique and the list of fundamental mitigation techniques suggested by the VAM methodology provides insight into how an AM security technique can be improved. Finally, the paper provides a case study of the proposed adapted VAM methodology to demonstrate its application.

Keywords: Industry 4.0, Evolutionary Computation, Optimization
Abstract: Batch scheduling problems deal with jobs to be processed in batches in many industrial production systems. They are hard to solve. This work proposes a novel bi-objective batch scheduling problem with the constraints of release time and sequence-dependent setup time. As an important characteristic of the concerned problem, the number of late jobs within a batch varies with its start time. A mixed-integer linear program is proposed to describe this problem. Two objectives, i.e., minimizing the total number of late jobs and setup time, are considered. Two memetic algorithms by integrating a nondominated sorting genetic algorithm II (NSGA-II) and 2-opt local search are designed to solve the concerned problem. They adopt different crossover operators, i.e., partially mapped one and precedence preserved one. By comparing the results of the proposed algorithms with their peers on extensive experiments, we conclude that the proposed algorithms get much better Pareto fronts than their peers at the expense of more execution time. Yet, their speeds are fast enough to solve the problems with industrial scales and thus prove the readiness to put them in industrial use.

Keywords: Industry 4.0, Machine Learning, Cybernetics for Informatics
Abstract: The information securities of industrial networks are getting crucial for the reliability of smart and connected industrial infrastructures and systems. Although different anomaly detection methods have been proved to be effective for industrial networks, the implementations are too slow to be used in practice. In this paper, a semi-supervisory anomaly detection method is proposed to upgrade the implementation efficiency as well as the detection accuracy. The proposed method, in the first-fold, executes a manifold learning technique to reduce arbitary network protocol data dimension down to two dimensions, then in the second-fold, K-NN classification is implemented to recognize abnormal network data. The results illustrate that the proposed method is superior to traditional anomaly detection methods on the aspects of both accuracy and time consumption.

Keywords: Industry 4.0, Neural Networks and their Applications, Hybrid models of NN
Abstract: The high-speed solenoid valve is a crucial module in train braking system, which is an essential factor to ensure the safe operation of trains. How to estimate the health status of the high-speed solenoid valve accurately to improve the reliability of train braking system is a challenging issue. Most related work relies on accurate physical models or large amounts of historical data. To address this challenge, this paper proposes a hybrid datafusion estimate method for the health status of train braking system. Firstly, the physical model of the high-speed solenoid valve is established, and physical indicators which represent the working performance are extracted. Then, the dynamic driving current is processed by ensemble empirical mode decomposition (EEMD) to calculate the information entropy. Physical indicators and information entropy indicators are combined into a feature vector, which can be reduced the dimension by the t-distributed stochastic neighbor embedding (T-SNE) algorithm. Finally, the feature vector is input into the probabilistic neural network (PNN) to estimate the health status of train braking system. The proposed method is implemented in the high-speed solenoid valve degradation dataset, which collected by the train brake system experiment platform. The result shows that it is better than other methods in the accuracy and calculation efficiency.

Keywords: Machine Learning, Neural Networks and their Applications
Abstract: The traditional gradient descent based optimization algorithms for neural network are subjected to too many vulnerabilities, such as slow convergent rate, gradient vanishing and falling into local minima. Therefore, the alternative non-gradient descent learning algorithm was proposed and prevalently applied in kinds of domains, such as pseudoinverse learning algorithm (PIL). However, when a special variant of the PIL, taking the random configuration of weight parameters, is adopted, the generalization ability needs further improvement although it has excellent training efficiency. Thus, on consideration of integrating the idea of ensemble learning, we propose two methods to enhance basic PIL. One method is equivalent to an additive model, which can raise the network's performance by introducing boosting mechanism, and the other is to adopt a recursive way to rectify the hidden layer output of the neural network, then the relative better model is used in the subsequent prediction. Comprehensive evaluating experiments are conducted on several datasets, and the experimental results illustrate that the our proposed methods are effective on the classification accuracy.

Keywords: Machine Learning, Neural Networks and their Applications
Abstract: In this paper, we construct the neural network that learns a fall motion by measuring a behavior that simulates a fall forward due to a trip, in order to realize a system to predict a fall a little ahead. Recent advances in machine learning techniques have enabled the development of methods for predicting behavior in real time. This is expected to be used for walking to predict a fall and support it in advance to reduce injuries. Although many systems have been proposed to measure and detect a fall, there are few studies on the data that a fall is caused by an unintentional disturbance during normal walking. Therefore, we do not know how long it takes for a person to fall over after a disturbance is given, and we do not have much understanding of how predictable the phenomenon is in principle. In this paper, we constructed a system to simulate a fall with a disturbance and measured the 3D skeletal data of the system. From these results, the average time between the disturbance and the start of the fall was calculated. By using a neural network to make predictions, we confirmed that falls can be predicted at that frame given the disturbance.

Keywords: Machine Learning, Neural Networks and their Applications
Abstract: The studies addressing the application of machine and deep learning models to analyze the sentiments of Arabic online reviews related to the real-estate and automobile fields are not mature. To fill this gap, this research has focused on classifying three types of sentiments in Arabic real-estate and automobile online reviews, which are negative, positive, and mixed sentiments. The research focused on analyzing the reviews written in both Gulf Cooperation Council (GCC) dialects and modern standard Arabic (MSA). The research also explained the natural language processing strategies that were adopted to prepare the text for classification. The research discussed the details of collecting and annotating the data, pre-processing procedures, and feature selection methods. Following this, the research highlighted the adopted strategies for balancing and splitting the datasets, and it showed the analysis of the classification results for both machine and deep learning models. Finally, the suggestions for future work were provided in this research.

Keywords: Machine Learning, Neural Networks and their Applications, Agent-Based Modeling
Abstract: Simulation-based training has the potential to significantly improve training value in the air combat domain. However, synthetic opponents must be controlled by high-quality behavior models, in order to exhibit human-like behavior. Building such models by hand is recognized as a very challenging task. In this work, we study how multi-agent deep reinforcement learning can be used to construct behavior models for synthetic pilots in air combat simulation. We empirically evaluate a number of approaches in two air combat scenarios, and demonstrate that curriculum learning is a promising approach for handling the high-dimensional state space of the air combat domain, and that multi-objective learning can produce synthetic agents with diverse characteristics, which can stimulate human pilots in training.

Keywords: Machine Learning, Neural Networks and their Applications, Cybernetics for Informatics
Abstract: Analyzing and detecting Border Gateway Protocol (BGP) anomalies are topics of great interest in cybersecurity. Various anomaly detection approaches such as time series and historical-based analysis, statistical validation, reachability checks, and machine learning have been applied to BGP datasets. In this paper, we use BGP update messages collected from Reseaux IP Europeens and Route Views to detect BGP anomalies caused by Slammer worm, WannaCrypt ransomware, and Moscow blackout by employing recurrent neural network machine learning algorithms.

Keywords: Neural Networks and their Applications, Machine Learning, Computational Intelligence
Abstract: In recent times, Social Network Analysis (SNA) has become a very important and interesting subject matter with regard to Artificial Intelligence (AI) in that a vast variety of processes, comprising animate and inanimate entities, can be examined by means of SNA. As a result, prediction tasks within social network structures have become significant research problems in SNA. Hidden facts and details about social network structures can be effectively and efficiently harnessed for training AI models with the goal of predicting missing links/ties within a given social network. Thus, important factors such as the individual attributes of spatial social actors, and the underlying patterns of relationship binding these social actors must be taken into consideration because these factors are relevant in understanding the nature and dynamics of a given social network structure. In this paper, we have proposed an interesting hybrid model: Representation Learning via Knowledge-Graph Embeddings and ConvNet (RLVECN). Our proposition herein is designed for examining, extracting, and learning meaningful facts for resolving link prediction problems about social network structures. RLVECN utilizes an edge sampling approach for exploiting the representations of a social graph, via learning the context of each actor with respect to its neighboring actors, with the goal of generating vector-space embeddings per actor which are further harnessed for innate representations via a Convolutional Neural Network (ConvNet) sublayer. Successively, these relatively low-dimensional representations are fed as input features to a downstream classifier for solving link prediction problems in a given social network. Our proposition, RLVECN, has been trained and evaluated on six (6) real-world benchmark social graph datasets.

Keywords: Neural Networks and their Applications, Machine Learning, Computational Intelligence
Abstract: Time series data is available from a diverse set of sensors in real life. It is of prime importance in the domain of machine learning and artificial intelligence to analyze such data and identify outliers or anomalies, characteristic of the underlying activities and predict the future. Traditionally, time-series analysis involves identifying features using exploratory data analysis and using statistical approaches for classification and prediction. However, with the advent of convolutional neural networks (CNN), our ability to extract features automatically has substantially improved. In this paper, we propose a novel light-weight deep learning architecture of dilated CNN for classification and predicting time series data sets. We evaluate our model on a real-world human activity recognition time series data set and a synthetically crafted pseudo-realistic dataset for human intent recognition. Our model outperforms the state-of-the-art models and is light-weight.

Keywords: Neural Networks and their Applications, Machine Learning, Computational Intelligence
Abstract: The driver of technology innovation is shifting from raw computing performance to performance delivered per watt. Therefore, it is crucial to conduct heterogeneous (CPU-GPU) system performance analysis in terms of power utilization. The main objective of our experimental study is to provide a detailed analysis of performance and power utilization of Convolution Neural Network for image classification of CIFAR-10 tiny images. We present an approach to calculate one convolution-layer power utilization for heterogeneous CPU-GPU systems by employing CUDA and OpenCL environments. The purpose of power, performance and hardware utilization analysis is to promote green computing and to assist system designers and AI specialists in choosing a green neural network architecture for energy-aware high-performance heterogeneous systems.

Keywords: Computational Intelligence, Swarm Intelligence, Evolutionary Computation
Abstract: Mutation Testing is an effective testing technique that relies in the generation of mutants from the system under test. The main limitation of this technique is that the potential number of mutants is usually huge. Therefore, it is important to classify and select mutants in order to avoid repetitive, useless or excessive computations, and biased results. In this paper we focus on avoiding too many executions and/or biased results by classifying mutants into two categories: hard-to-kill and easy-to-kill mutants. We propose a new swarm intelligence algorithm to classify a set of mutants between those two classes and we show how our algorithm compares to other approaches.

Keywords: Agent-Based Modeling
Abstract: Knowledge diffusion and opinion formation are important research topics in the area of communication in social networks. We present our generic model with interchangeable variants for processing messages from other members of the group. We describe several generic aspects of communication, like learning, persuasion, or polarization, and apply them in our model. We then show how this simple change leads to different results for the communication process of the whole group. We show how the model with interchangeable integration strategies is applicable for modeling various aspects of online social network and real life communication.

Keywords: Fuzzy Systems and Evolutionary Computing, Heuristic Algorithms, Optimization
Abstract: Discrepancies arise when experts have to decide the preference on alternatives for a problem. Henceforth, it is necessary to carry out a process during which they adjust their opinions in order to achieve an acceptable consensus. Usually, this process is coordinated by a moderator that makes suggestions to the participants regarding the most adequate changes. In order to simplify this process, we propose an evolutionary technique based on the search of the changes of the preferences that improve the consensus degree. We also consider that the opinions of the experts should stay closer to the original ones. Taking into account these two factors, we are able to provide useful feedback for the experts willing to get a consensus and measure how much improvement can be achieved.

Keywords: Machine Learning
Abstract: The paper presents a method for creating homogeneous and heterogeneous ensembles using three measures of regressor diversity, namely chi-square statistic, correlation coefficient and disagreement. Regressors included in the ensemble were selected from a pool of 100, 200 and 300 models generated using algorithms included in the WEKA data mining software package, such as M5P, M5Rules, Random Tree, and REPTree. Using three different measures, the degree of diversity of each pair of regressors was determined, and then 40 models with the highest diversity value were selected to construct ensembles. The obtained regressor ensembles were compared in terms of accuracy of prediction with models created using standard machine learning techniques such as linear regression, k-nearest neighbors, bagging and random forest. For comparative experiments, a real data set of purchase and sale transactions of residential real estate taken from a cadastral system was used. The research showed the usefulness of the proposed method for building ensembles.

Keywords: Companion Technologies, Human Factors, Entertainment Engineering
Abstract: Smart toys raises new concerns for parents and researchers. Children are more likely to share sensitive data and are unaware or rarely care about online risks. Parents play a relevant role in protecting the children, and parental control tools are necessary to take control and properly manage their child's data, according to their preferences. However, current tools neither meet parental needs nor are compliant with a standard for toy makers. We present a study of requirements for the development of a parental control tool for smart toys.

Keywords: Human Factors, Augmented Cognition, Assistive Technology
Abstract: This paper addresses the differences between conventional optical mirrors and electrical mirrors consisting of a camera and a flat display panel. Recently, vehicle mirrors (side-view and rear-view mirrors) are gradually being replaced with electrical mirrors (i.e., display devices). The characteristics of electrical mirrors are different from those of conventional mirrors in many aspects. These differences can cause an uncomfortable feeling in drivers. In this paper, we focus on the half occlusion appearing in the peripheral areas of a display and investigate the effects of half-occluded regions on whether humans recognize a viewing device as a mirror. To evaluate this effect, we conducted experiments based on pairwise comparisons using a head-mounted display (HMD) to present virtual mirror objects having different characteristics in terms of half occlusion as well as binocular and motion parallaxes. Experimental results suggest that half occlusion is certainly related to the human recognition of a mirror. Furthermore, pseudo half-occlusions introduced by barriers in front of a 2D display can reinforce this recognition.

Keywords: Human Factors, Human-Machine Cooperation and Systems, Human-Machine Interface
Abstract: We conducted two experiments to examine the effect of providing automation support and communicating the limitations of the automation on team performance in a simulated navy environment. Two-person teams engaged with a picture compilation task with or without automation support. In the first experiment, there was no explicit explanations provided regarding the automation’s limitations. In the second experiment, limitations of the automation were communicated to the participants verbally. A comparison of two experiments revealed that participants classified fewer contacts when the automation support was present. Moreover, communicating the limitations of automation resulted in even fewer classifications than when no information was provided. Possible reasons for these results include confusion created by the additional information and reprioritization. These results highlight the complexity of delivering automation transparency to operators in safety-critical environments. We conclude that automation transparency should be carefully designed and delivered to avoid negative impacts on performance.

Keywords: Human-Machine Interface, Human Factors, Human-Computer Interaction
Abstract: Critics of Ecological Interface Design (EID) laboratory studies have argued that the experiments are confounded by non-equivalent sensor sets reflected in the contrasting interfaces. Whereas some EID proponents have responded by emphasizing theoretical justifications for the interfaces employed, the critique is grounded in practical challenges that endure in human-machine interface research. We construct an experiment from data that includes an interface that removes the sensor set confound. The results are surprising. Adding additional sensor information to the data-impoverished non-EID interface results in poorer task performance and no improvements in control stability. Participants using the EID interface exhibited superior task performance and control stability to those in two non-EID conditions. This result speaks to the generalizability of EID research findings to industry applications.

Keywords: Human-Machine Interface, Human Factors, Human-Computer Interaction
Abstract: Situational Awareness (SA) is paramount to ensuring the Nuclear Power Plant and Commercial aviation industry are operated safely. An increase in Human-in-the-loop (HITL) error rate may be indicative of challenged operator SA while undermining safety. In this paper, a natural language processing approach for modeling industrial Human Machine Interface (HMI) state transitions is introduced towards detecting operator HITL error precursors in real-time. A custom language translation seq2seq Encode-Decoder model design is implemented and evaluated using a real-plant scenario dataset obtained from a Nuclear Power Plant Operator training simulator. Results support NLP HMI state model can detect HITL error precursor within N time-steps prior to an accident event.

Keywords: Human Performance Modeling, Human Factors, Human-Computer Interaction
Abstract: Drowsiness at the wheel is a major problem for traffic road safety. A drowsy driver suffers from decreased vigilance, increased reaction time and degraded decision-making ability, all of which have a huge impact on the driving performance. A driver monitoring system that warns the driver of his or her critical drowsiness state is a worthwhile contribution to traffic road safety. A drowsy driver typically exhibits some observable behaviors, such as eye blinking and head movements, that can be tracked using a camera. In this study, we analyze the potential of eye closure and head rotation signals, provided by a driver camera, to classify the driver's drowsiness state using logistic regression models. This analysis is based on a large dataset collected from 71 subjects in driving simulator experiments. A reliable and independent reference for drowsiness, however, is required in order to perform this analysis. For this purpose, we devise a methodology that merges several drowsiness monitoring approaches to construct a reliable reference for drowsiness. Furthermore, we describe our approach to extract eye blink and head rotation features. Ultimately, we design logistic regression classifiers and combine them using the one-vs-one binarization technique. Our approach achieves a global balanced validation accuracy of 72.7% on a three-class classification problem (awake, questionable and drowsy) by adopting a strict and rigorous evaluation scheme (i.e., leave-one-drive-out cross-validation).

Keywords: Human Performance Modeling, Human Factors, Human-Machine Cooperation and Systems
Abstract: Humans skillfully operate many types of advanced human-robot/machine systems, such as automobiles, using their upper and/or lower limbs. Some researchers have proposed methodologies for integrating biomechanical/perceptual models of human operators into such systems to improve their safety and maneuverability. However, there has been no research focusing on the biomechanical/psychological interactions between upper and lower limbs, even though human operators often have to utilize both sets of limbs simultaneously. Thus, this study aimed to investigate human force perception properties (HFPP) with consideration for the perceptual interactions between upper and lower limbs during the operation of vehicular driving devices, such as a steering wheel and gas/brake pedal. A set of force perception experiments was conducted on eight healthy subjects. The subjects first determined the standard force magnitude applied to their right foot by the pedal, then determined the magnitude of steering force when using one or two hands, which was equivalent to the standard force magnitude. Experimental results demonstrate that the HFPPs for a single hand and both hands change with the standard force magnitude and that the HFPP for a single hand is significantly influenced by steering rotational directions resulting in smaller standard force magnitudes. Furthermore, a computational model of HFPPs for both hands was developed to explain the perceptual interactions between upper and lower limbs.

Keywords: Human Performance Modeling, Human Factors, Human-Machine Cooperation and Systems
Abstract: This study aims for the new driving skill improvement support system based on driving skill evaluation. We try to define driving skill in terms of "spatial perception" and "prediction". This research can be roughly divided into two steps. The first step is to establish an objective skill evaluation model. The second step is to verify effects of the support system for driving skill improvement. In this paper, we describe that we evaluated driving skill based on parking performance, operational behavior like steering and cognitive behavior like head motion and gazing points. We conducted driving behavior measurement experiment by using a real vehicle to get operational and cognitive data during driving. In consequence, we confirmed that driving skill could be defined by spatial perception and operational behavior of drivers in approaching to an obstacle.

Keywords: Human Performance Modeling, Human Factors, Information Visualization
Abstract: Understanding the relationship between driving behavior and visual information is an important issue in order to understand driving behavior holistically. In this study, we constructed a driver model that reproduces the driver's steering behavior from visual information based on the Convolutional Neural Network (CNN) with human physical characteristics. We obtained the driving behavior in a simulator study to train the proposed CNN model. Which region in the visual field influencing drivers' steering behavior was analyzed using the results of the feature maps generated by the trained CNN model and the driver's gaze behavior. The results indicate that the drivers perform steering action using the information within 20 degrees from the gaze point.

Keywords: Human Performance Modeling, Human-Machine Cooperation and Systems
Abstract: A new concept for modeling the behavior and the interaction between drivers at intersections is presented. For this purpose a hybrid dynamic game is introduced using the concepts of game theory. The proposed hybrid game is used to model the interactive behavior in a junction scenario. The presented hybrid approach divides the modeling of motion into individual maneuvers. With this partition of motion, the decision process can be reduced to a simple maneuver selection that takes into account the motion of the vehicle without the need for solving complex coupled differential equations. In order to make computation of a solution feasible we propose a rule based adaption mechanism. Simulation is used to show the applicability of the developed hybrid dynamic game approach.

Keywords: Human Performance Modeling, Affective Computing, Kansel (sense/emotion) Engineering
Abstract: Head pose offers vital cues to infer one's social attention in wide applications. Most existing head pose estimation algorithms have demonstrated competitive results taking high resolution images of frontal view as input. However, these approaches still work poorly if they are fed with low-resolution images. In a more common realistic scene such as computer vision assisted autism screening, images of unconstrained patients that are taken from distant cameras often have low-resolution and non-frontal faces. To this end, we present a multi-view scheme based CNN for free-head pose classification. Residual Networks are taken as the backbone model to generate effective feature representations of the low-resolution images. A novel multi-view feature fusion layer is proposed to address facial appearance variation over multi perspectives owing to free movement. Also, a multi loss function by combining binned classification and regression losses of different pose angles is employed to obtain a more precise pose estimation. The proposed method is evaluated on two scenarios: (1) a publicly available dataset and (2) a practical application to explore the social attention of a group with social deficits: autistic children. Experimental results suggest that our method significantly outperforms state-of-the-art multi-view pose classification methods and achieves comparable pose estimation results. Moreover, the proposed method can be extended to applications of quantitative analysis of social deficits.

Keywords: Human Performance Modeling, Human-Machine Cooperation and Systems
Abstract: Driver modeling is essential in the development of haptic guidance systems. A new cybernetic driver model designed to account for the cooperation between the driver and haptic guidance systems has recently been proposed. This paper aims to validate this model in situations of interaction with different levels of haptic guidance on a driving simulator. Two experiments have been performed for this purpose. The first experiment consisted of implementing the driver model in the driving simulator and evaluating its lateral control performance when interacting with a haptic guidance system. The results reveal that the model can be adapted to different sharing levels by adjusting only the gain of an internal model of the steering wheel compliance. The second experiment consisted of estimating the evolution of the gain of this internal model using the unscented Kalman filter. The results reveal consistency between the evolution of the identified parameter and the level of sharing of the haptic guidance system. The driver model represents the process of human driver adaptation to variations in the level of sharing in haptic guidance systems.

Keywords: Human Performance Modeling, Human-Machine Cooperation and Systems, Human-Machine Interface
Abstract: In the past decade, remarkable research has been done on human-machine cooperation to generate synergies and mutual benefits. However, most research so far only considers the control level of interaction with concepts like haptic shared control. This paper focuses on the emerging research on human-machine cooperation on higher levels of interaction to tackle more complex challenges. Therefore, we first introduce a generalized level model based on established models to define our research emphasis on emancipated human-machine cooperation on all levels. Second, the design and results of a study on human-machine cooperation on decision level are presented. We examine the negotiation behavior of humans in a scenario with discrete decision options and a deadline. The results indicate the validity of a previously proposed model based on negotiation theory to describe the observed human behavior. Additionally, the observed influencing factors on the negotiation behavior are crucial for a proper automation design: adaptation and identification methods are required to enable the automation to take part in an emancipated negotiation with a human.

Keywords: Human-Computer Interaction, Human Factors, Assistive Technology
Abstract: The screening and diagnosis of Autism Spectrum Disorder(ASD) suffer from great challenges due to insufficient professional clinicians and complex procedures. It is urgent to introduce an effective auxiliary system in the diagnosis and treatment process to assist in the completion of pathological information collection tasks, consequently simplifying the screening method and improving the accuracy of screening. We propose a computer vision-based early screening system for ASD to characterize the facial expressions and eye gaze attention which are considered to remarkable indicators for early screening of autism. The system provides the subjects with three different virtual interaction modes: video, picture, and virtual interactive game. During the interaction between the subject and the computer, the system extracts and analyzes the quantitative information of the subject's performance. Then, through computer vision-based emotion analysis and attention analysis methods, the subject's emotions and attention features in the three interaction modes are automatically calculated to assist in the early screening of autism. Finally, the accuracy and feasibility of the system are verified through experiments on both the publicly available dataset and the data collected from 10 ASD children.

Keywords: Wearable Computing, Brain-based Information Communications, Human-Computer Interaction
Abstract: Electroencephalography (EEG) headsets are wearable computing devices capable of recording electrical activity of the brain. These devices play a key role in the Brain-Computer Interfaces (BCIs) systems, i.e., systems capable of acquiring, processing and classifying EEG signals in order to control external devices such as wireless prosthetics. In spite of their crucial role, the current EEG headsets are very uncomfortable being composed of many wet electrodes. Hence, single-channel BCIs with dry electrodes are emerging like wearable devices more accepted by users. Unfortunately, this kind of device typically provides weaker and noisier signal that makes more challenging the classification task. This work is aimed at improving the quality of the classification of EEG signals, and in particular of Steady-State Visual Evoked Potentials (SSVEP), captured by single-channel EEG devices by using an evolutionary algorithm-based optimized version of Support Vector Machine (SVM). As shown by experimental results, the proposed approach improves on the state-of-the-art methods in terms of accuracy.

Keywords: Intelligent Power and Energy Systems, Infrastructure Systems and Services
Abstract: A transmission power system carries the energy produced by generators to load points over long distances and to ensure the power demand is satisfactorily reached, the electric utilities use devices running applications for monitoring, control and protection, in order to maintain system operation in a normal state. These applications mainly require continuous data communication and low latency, which makes crucial to define the appropriate communication technology. In this context, this paper proposes a study which simulates a scenario based on the IEEE 14-bus transmission test system, on NS-3 (Network Simulator 3) tool, using LTE and WiMAX wireless technologies to move data across data links on the system. The simulations generate metrics like delay, throughput and packet loss, to compare the technologies and express the degree of compatibility of each one of them with the applications tested, which can guide the utilities in choosing the communication technology to be deployed in its system.

Keywords: Intelligent Power and Energy Systems, Intelligent transportation systems
Abstract: Different from previous data-driven methods for lithium-ion battery State-of-Charge (SoC) estimation, this paper aims to develop a hierarchical SoC estimation method to address the data dependency issue and measurement noise interferences. In the off-line training layer, aging-aware features are extracted to improve SoC estimation accuracy throughout the entire battery life cycle. Extreme gradient boosting (XGBoost) is introduced to map the relationship between the extracted features and SoC for its strong nonlinear fitting ability. In the on-line estimation layer, Ampere-hour integral method is utilized to provide SoC reference to guarantee the stability of the proposed method. Meanwhile, to suppress the measurement noise, we adopt Kalman filter to correct the SoC value estimated by XGBoost. The superiority of the proposed method is proved under the random walk discharging experiment by comparing with the results of XGBoost, i.e., without Kalman filter. The proposed method improved the accuracy of lithium-ion battery SoC by 4% to 10%.

Keywords: Intelligent Power and Energy Systems, Decision Support Systems, Model-based Systems Engineering
Abstract: Quickly identifying faulty sections is tremendously important for power systems, yet challenging due to handling the variations of complex alarm patterns. Existing works have focused on finding fault section clues solely from alarm information (and ignoring power system topology information). So they are only sensitive to alarms from power systems with pre-assumed topology structures, and encounter difficulties when a system’s topology changes. To adapt to unknown or varying system topologies, here we present a Topology-Adaptive Deep Model (TADM) for power system multifault diagnosis. TADM mines the underlying mapping from alarm and topology information to each section’s fault status. It consists of a deep iterative network (DIN), a one-layer fully connected network (FCN), and section-wise multifault diagnosis (SWMD) subnetwork. TADM first models a fault power system as a graph, from which DIN iteratively integrates the alarm and topology information in the region from each node to its T-hop neighbors, and learns their local correlation. Limited to T’s size, FCN then combines all local correlations to determine the global correlation between alarm and topology information across the entire power system. To implement multifault diagnosis, learned local and global correlations serve as topology-related fault representations for input as an SWMD (to predict all sections’ fault states one by one). A comprehensive experimental study demonstrates that TADM outperforms state-of-the-art models in both multifault diagnosis and adapting to system topologies. The source code of the TADM is available onlline1.

Keywords: Model-based Systems Engineering, Intelligent Power and Energy Systems, Infrastructure Systems and Services
Abstract: Pipe bursts and leaks in district heating networks are a problem both for the economic operation and for the supply reliability of the connected consumers. In case of a leakage, pressure and flow rate conditions near the defect change. These changes spread to the entire network within a short period of time and, depending on the size of the leakage, partly lead to drastically changed network conditions. After a leakage is detected, it is necessary to localize the leakage as accurate as possible in order to shut down the affected network segment and maintain the network's stability. This article discusses and compares three different approaches for leakage localization (pressure wave detection, model-based numeric-analytical and machine learning) that exploit different properties of simulation models and sensor information from the real network.

Keywords: System of Systems, Model-based Systems Engineering, Cyber-Physical Cloud Systems
Abstract: With the continuous growth in wire density, the reliability has become a dominant burden while channels of a modern NoC are exposed to various faults. A key requirement for the NoC is therefore to propose a mechanism that can account for a channel fault since it significantly impacts NoC performance. This paper presents a distributed test strategy that detects and diagnoses logic-level faults coexist in NoC channels and deeply analyze the severe impact of these faults on network performance. Fault coexistence in channels makes a fraction undetectable and is addressed here. Simulation results demonstrate the effectiveness of the proposed strategy.

Keywords: Intelligent Learning in Control Systems, Model-based Systems Engineering, Recommender Systems
Abstract: Quantum computing and deep learning methods hold great promise to open up a new era of computing and have been receiving significant attention recently. This paper presents quantum computing (QC) based deep learning methods for fault diagnosis that are capable of overcoming the computational challenges faced by conventional techniques performed on classical computers. The shortcomings of such classical data-driven techniques are addressed by the proposed QC-based fault diagnosis model. A quantum computing assisted generative training process followed by supervised discriminative training is used to train this model. The applicability of proposed model and methods is demonstrated by applying them to process monitoring of Tennessee Eastman (TE) process. The proposed QC-based deep learning approach enjoys superior performance with an average fault diagnosis rate of 80% and tremendously low false alarm rates for the TE process.

Keywords: Intelligent Learning in Control Systems, Decision Support Systems, System of Systems
Abstract: In this study, an efficient strategy for fault detection and isolation (FDI) of an Industrial Gas Turbine is introduced based on ensemble learning methods. Four independent Wiener models are identified by employing plant input/output data to determine system behavior. Following that, an ensemble-based method is established, which utilizes all the Wiener models and relevant residuals to detect the faults. A fault isolation structure is then developed based on ensemble bagged tree procedure such that it is capable of isolating faults in a steady-state runtime. As a crucial goal, increasing accuracy and robustness simultaneously are mainly centered. The proposed FDI method is tested on nonlinear gas turbine simulation using real data from a combined cycle power plant. The obtained results illustrate the correctness and accuracy of the presented FDI scheme.

Keywords: Intelligent Learning in Control Systems
Abstract: This paper investigates a new recursive least squares (RLS) algorithm for active control (AC) of impulsive noise sources. The proposed RLS algorithm is based on an objective function having robustness against the impulsive noise. This paper introduces two versions of the proposed algorithm. In the first version, the update equation comprises a fixed step-size to regulate the adaptation. Considering that a fix-valued step-size results in a trade-off situation from the view-point of the performance during the transient and at the steady-state, a second version of proposed algorithm addresses this trade-off by incorporating a convex combined step-size (CCSS) strategy. The CCSS strategy selects (automatically) a large-valued step-size at the start-up for a fast convergence speed. As the AC system approaches the steady-state, the CCSS is adapted to a small value for improving the performance at the steady-state. The simulation results show that the proposed algorithm is very effective in the various settings of AC systems for impulsive sources.

Keywords: Intelligent Learning in Control Systems
Abstract: A data-driven computational heuristic is proposed to control MIMO systems without prior knowledge of their dynamics. The heuristic is illustrated on a two-input two-output balance system. It integrates a self-adjusting nonlinear threshold accepting heuristic with a neural network to compromise between the desired transient and steady state characteristics of the system while optimizing a dynamic cost function. The heuristic decides on the control gains of multiple interacting PID control loops. The neural network is trained upon optimizing a weighted-derivative like objective cost function. The performance of the developed mechanism is compared with another controller that employs a combined PID-Riccati approach. One of the salient features of the proposed control schemes is that they do not require prior knowledge of the system dynamics. However, they depend on a known region of stability for the control gains to be used as a search space by the optimization algorithm. The control mechanism is validated using different optimization criteria which address different design requirements.

Keywords: Recommender Systems, Intelligent Learning in Control Systems, Model-based Systems Engineering
Abstract: Recommender system (RS) commonly describes its user-item preferences with a high-dimensional and sparse (HiDS) matrix. A latent factor (LF) model relying on stochastic gradient descent (SGD) is frequently adopted to extract useful information from such an HiDS matrix. In spite of its efficiency, an SGD-based LF model commonly takes many iterations to converge. When processing a large-scale HiDS matrix, its computational efficiency should be further improved by further accelerating its convergence rate as well as maintaining its learning ability. To address this issue, this paper innovatively proposes novel SGD algorithm which incorporates a nonlinear proportional integral derivative (NPID) controller into its learning scheme for building an LF model. The main idea is to adopt an NPID controller to model the learning residual achieved in the past iterations, thereby adjusting the learning direction and step size of the current iteration, thereby making a resultant model converge fast. With the NPID-incorporated SGD algorithm, this study proposes an NPID-SGD-based LF (NSLF) model. Experimental results on two HiDS matrices demonstrate that compared with a standard SGD-based LF model, the proposed model achieves higher computational efficiency and prediction accuracy for missing data of an HiDS matrix.

Keywords: Cooperative Systems, Cyber-Physical Cloud Systems, Robotic Systems
Abstract: This paper proposes a protocol that ensures trust between two vehicles in a multi-vehicle system. Trust is the implicit assessment that another vehicle will follow a predetermined strategy. The communication is done through a channel and the quantity of information transferred is guaranteed to be small. For privacy, the channel can be encrypted, but the message can only be decoded if the vehicles know the control strategy being followed. The protocol is implemented for a problem of two Unmanned Aerial Vehicles (UAVs) trying to find a target in a maze. The control strategy is implemented using Markov Decision Processes (MDPs). Simulations of the protocol demonstrate that communication is received and decoded by the teammates without explicitly revealing the tactics being used.

Keywords: Robotic Systems, Cooperative Systems
Abstract: The problem of a networked uncertain EulerLagrange systems (followers) to track a virtual dynamic leader under asymmetric time-varying communication delays is studied in this paper. It is assumed that the network is a directed spanning tree with the virtual leader as the root. Due to highly nonlinear of Euler-Lagrange systems and communication delays, it is challenging to design a control algorithm for followers to track the moving leader. To cope with the problems, we proposed a distributed cascade control framework which decouples an estimate of the leader velocity in the task space and an adaptive controller in the generalized space. It is verified that the network asymptotically achieves task-space consensus. Simulation results of networked Omni-directional mobile robots are provided to demonstrate the efficacy of the proposed control algorithm.

Keywords: Robotic Systems, Cooperative Systems
Abstract: To meet the demand for high additive manufacturing efficiency, we are working on an FDM (Fused Deposition Modeling) printer called DEXTER which has dual Selective Compliance Articulated Robot Arms (SCARA). We develop a path planner to produce time-optimal motion paths for each of DEXTER’s arms while guaranteeing that the arms do not collide with one another. We present a collision-free path planner for DEXTER’s arms using an improved Sampling-Based Model Predictive Optimization (SBMPO) based on A* type optimization by adding efficient collision determination and a new type of cost function. The simulation results show that the improved SBMPO can be used to efficiently generate smooth collision-free paths and trajectories with bounded velocity and acceleration.

Keywords: Robotic Systems
Abstract: Robots have inherently limited onboard processing, storage, and power capabilities. Cloud computing resources have the potential to provide significant advantages for robots in many applications. However, to make use of these resources, frameworks must be developed that facilitate robot interactions with cloud services. In this paper, we propose a cloud-based architecture called Smart Cloud that intends to overcome the physical limitations of single- or multi-robot systems through massively parallel computation, provided on demand by cloud services. Smart Cloud is implemented on Amazon Web Services (AWS) and available for robots running on the Robot Operating System (ROS) and on the non-ROS systems. Smart Cloud features a first-of-its-kind architecture that incorporates JavaScript-based libraries to run various robotic applications related to machine learning and other methods. This paper presents the architecture and its performance in terms of CPU usage and latency, and finally validates it for navigation and machine learning applications.

Keywords: Robotic Systems, System of Systems
Abstract: This paper reports on work to assess the feasibility of gravity-based long range underwater navigation and localization. As a first step, this is explored in simulations with RAO-Blackwellized particle filter simultaneous localization and mapping (SLAM). When implemented on an autonomous underwater vehicle it can operate submerged for extended periods without the use of an active sensor, thus widening the variety of AUV missions. Additionally, this work applies information theory to navigate through a region such that the SLAM data association, and thus the localization, performance is improved. The results also indicate that characteristic values for a region can be used as a SLAM metric for the region. Combining the characteristic value with information theory techniques improves the localization performance at extended ranges and is a first step towards long range underwater localization using gravimeters. Future work will optimize the particle filter, explore more sophisticated loop closures as well as hardware-in-the loop tests.

Keywords: Decision Support Systems, Enterprise Architecture & Engineering, Smart urban Environments
Abstract: The value of real housing property is affected by its geographical location. The problem consisted of estimating parameters that give importance to new investment in some districts, enabling to assess the best district of São Paulo to invest in new real estate developments. Thus, it was decided to use a linear regression hierarchical model, with random slopes per district in a binary variable that indicates new apartments. The objective of this work was to show how hierarchical models can be used to help the selection of the best place to develop a new real estate investment. The Bayesian inference was performed with Markov-Chain Monte-Carlo (MCMC), implemented in the JAGS R library. The results show that we can make pair-wise comparisons of the best locations for new apartments observing their posterior predictive probabilities of prices.

Keywords: Decision Support Systems, Intelligent transportation systems, Intelligent Learning in Control Systems
Abstract: The quayside running conditions play a key role in container terminal logistics systems, and the terminal liner berthing time (LBT) is the central index of quayside service efficiency that is also the important evidence and guidance to task scheduling and resource allocation at container terminals. The computational logistics and deep learning are combined to discuss the prediction of LBT by the generalization, unification and integration of the essence and connotation of computation. It is supposed to integrate the deep neural networks learning computation and logistics generalized computation for container terminals (LGC-CT) cross the boundaries between information space and physical world. A deep learning model is designed and executed to predict and evaluate LBT at a typical container terminal in China based on its LBT data for the past four years, which is also intended to lay a good foundation for the configuration, deployment and execution of LGC-CT. The deep neural networks are designed and implemented by the fusion of long short-term memory network, gated recurrent unit one, Gaussian noise one and dense one with TensorFlow 2.3, which demonstrates the feasibility and credibility of the proposed compound computing architecture and paradigms preliminarily.

Keywords: Smart urban Environments, Intelligent transportation systems, Model-based Systems Engineering
Abstract: This paper focused on the estimation of bus ridership using Wi-Fi probes (i.e., signals) emitted by smartphones that bus passengers carried. Smart stations -- which consist of a Raspberry-Pi computer, a Wi-Fi adapter, and a GPS add-on-- were programmed to sniff Wi-Fi signals and transmit signal data through a cloud service to the research computer. These smart stations were mounted onboard a network of transit buses that serve the City of Bozeman, Montana, and its surrounding areas. Two rule-based methods were developed to estimate the number of passengers onboard a bus at any given time. In the first, standard method, a signal was labeled as a passenger if it met arbitrary cutoff values from six criteria pertinent to speed (how fast the signal/device was traveling relative to the bus), duration being detected (as a proxy for how long the device remained in close proximity of the bus), and signal strength (which may correlate with the distance between the device and the bus). The second method employed a cost-function minimization via grid-search to tune the cutoff values involved in those subjective rules (e.g., a valid passenger signal should be close enough to a bus stop when it is first and last detected, but how close is close?). Results suggest a strong linear relationship between model-estimates and ground-truth passenger counts -- on average, the model estimates were able to capture 67% of the observed passenger counts. As Wi-Fi enabled personal devices continue to saturate the market, a Wi-Fi based counting tool as studied here can serve as an efficient way to monitor passenger flows of transportation systems.

Keywords: Intelligent transportation systems, Decision Support Systems
Abstract: This paper proposes an optimal car-following strategy for pure electric vehicles (EVs) with the aim of keeping an expected headway of the leader and reducing vehicle battery loss. In particular, a car-following system model is established. The primary task of the automatic vehicle is to follow the trajectory of the preceding car and maintain an expected headway. Then, the paper analyzes the powertrain of the electric vehicle. The loss of battery life over a period of time is proportional to the acceleration, so it takes the battery life into consideration. The Q-learning algorithm is conducted for the optimal car-following strategy using system data instead of system dynamics information. It utilizes reward function and greedy strategy to select actions to train the following vehicle to achieve car-following safety. When there is no collision in these two cars, acceleration is considered into reward function to reduce battery loss. Finally, it is verified by simulation that the proposed car-following strategy can keep good tracking, maintain the expected headway from the preceding vehicle, and reduce battery loss.

Keywords: Intelligent Assistants and Advisory Systems, Intelligent Learning in Control Systems, Intelligent transportation systems
Abstract: In subway stations, elevators are one of the most narrowed areas that slow down the moving of crowds. A large number of passengers gather around the elevator entrances and may cause unexpected accidents such as stampede. An effective way to guide the flow of passengers is to use guardrails. So far, the arrangement of guardrails in most subway stations is still designed manually, which requires rich experience and expert knowledge. In this paper, we propose to use the multi-bjective evolutionary algorithm to design the guardrails of the elevator entrance automatically. The transfer time of passengers and the flow rate are optimized concurrently. The proposed algorithm is tested in two scenarios with different complexities. Experimental results show that the proposed algorithm can provide promising guardrail arrangements, and reveal some instructive conclusions for guardrail design in subway stations.

Keywords: Systems Medicine, Model-based Systems Engineering, Intelligent transportation systems
Abstract: This research presented a new segmentation method based on fuzzy set, entropy and image fusion to analyze brain tumors from magnetic resonance imaging (MRI). Using fuzzy set, one can tackle the problem of uncertainty representation in gray levels of MRIs during the segmentation process. This uncertainty in their gray levels occurred due to poor illumination of images. To resolve this issue, this study focused on fuzzification of gray levels and assignment of membership degrees based on membership functions. Each fuzzified gray level value was quantified using entropy. The proposed method generated multiple thresholds based on maximum entropy values of gray levels. These thresholds generated multiple segmented images with different features. Finally, image fusion operation was performed on multiple segmented images to highlight all the critical features of brain tumors. Fusion images were compared with the segmented images obtained from four additional methods, the multilevel threshold method, adaptive threshold method, K-means clustering algorithm and fuzzy c-means algorithm. The performance evaluation metrics indicated the effectiveness of the proposed method over these existing methods.

Keywords: Model-based Systems Engineering
Abstract: Automatic real-time emotion recognition based on multi-channel EEG signals is a significant and challenging task in neurology and psychiatry. In recent years, deep learning has been used in EEG emotion recognition. However, many existing deep learning based methods still require complex pre-processing or additional feature extraction, which make it difficult to achieve real-time emotion recognition. In this paper, an end-to-end model named Temporal Convolutional Broad Learning System (TCBLS) was designed for multi-channel EEG based emotion recognition. The TCBLS takes one-dimensional EEG signals as input, then extracts emotion-related features of EEG automatically. In this model, the Temporal Convolutional Network (TCN) is designed to extract EEG temporal features and deep abstract features simultaneously, then Broad Learning System (BLS) is used to map the features to a more discriminative space and further enhance the features. We evaluated our method on DEAP database, performing 10-fold cross-validation on each subject to obtain the classification accuracy. Experimental results indicate that the performance of TCBLS is better than other comparison methods, and the mean accuracy of TCBLS is 99.5755% and 99.5781% on valence and arousal classification task respectively. The results demonstrate the effectiveness and robustness of TCBLS in EEG emotion recognition.

Keywords: Intelligent Assistants and Advisory Systems, Decision Support Systems
Abstract: This paper addresses the challenge of detecting the breathing cessation in sleeping subjects, via breathing pattern monitoring at a distance and under ''night-light'' conditions. We investigate a near-infrared video-based approach to estimate the breathing rate, based on chest or back movements. A body pose estimation algorithm and the Lucas-Kanade optical flow method are combined to automatically detect the Region of Interest (ROI) represented by a grid of points. The movement of the ROI is then translated into the frequency of respiratory events. We used a dataset with 28 near-infrared videos, as well as 11 videos of subject uncovered and partially covered by blankets. We compared the breathing rate measurements provided by a wearable device with the ones estimated by the video-based approach. A linear correlation analysis of both measurements resulted in a coefficient of determination of 0.925, and accuracy of 99.70% for the first dataset, and 0.873 and 88.95% for the second dataset, respectively. The ultimate application is to detect abnormalities in breathing and health emergencies in environments such as homeless shelters.

Keywords: Robotic Systems, Medical Mechatronics, Bio-mechatronics and Bio-robotics Systems
Abstract: Design, control and evaluation of a lower limb wearable robotic exoskeleton for power assistance are presented in the paper. The proposed four degree-of-freedom robotic exoskeleton, an active flexion/extension and a passive abduction/adduction rotation at each hip joint, is characterized with complying with the swinging motion of lower limbs as close as possible. To perform power assistance on walking, the linear extended state observer (LESO) based controllers were designed for the walking assistance. Finally, the experiments were conducted to validate the prototype of lower limb robotic exoskeleton. The associated evaluations for the walking assistance were also investigated using the motion captured system and EMG signal.

Keywords: Model-based Systems Engineering, Decision Support Systems, Intelligent Learning in Control Systems
Abstract: This paper presents an effective way to extract low-level features based on sparse coding for facial occlusion detection. Masked Neural Sparse Encoder (MNSE) is proposed to be a sparse coding solver that brings out better feature bases for data representation and improvement in the anomaly detection task. To guarantee the representational capability of features, a set of masks is applied to force each feature basis is heeded on learning a specific stroke within a certain area. The mask set is constructed by clustering primary strokes from training samples and represents them with corresponding centers of clusters. Hence, these masks stand for main strokes in concerned areas with higher probabilities. Experiments show MNSE contains better representational capability in data from different domains. Compared with the standard sparse coding and the auto-encoder based approaches, MNSE lifts the accuracy up around 20%.

Keywords: Brain-based Information Communications, Human-Computer Interaction, Human Performance Modeling
Abstract: This study aimed to evaluate whether current electroencephalographic spectral measures can predict participant's performance during future sessions of a motor imagery task. By investigating this point, we hope to understand which spectral components are related to MI "literacy". Twelve healthy subjects performed a neurofeedback task whereby a cursor was moved to one of two targets (left and right) using only motor imagery of the corresponding hands. To evaluate the effect of aptitude, we measured the Mahalanobis' distances between whole-scalp spectral patterns in four frequency bands (theta, alpha, beta, and gamma) during the first session of left and right motor imagery. Later, we used these features as inputs in a Support Vector Regressor to predict performance during the following two sessions. The performance was measured as the percentage of trials where the cursor correctly reached the target. Since our sample was balanced, this approach predicted performance on sessions two and three with mean absolute errors of 15.07±12.94% and 11.98±11.40%, respectively. The most relevant feature in both cases was the Mahalanobis' distance in alpha. These results suggest that participants who can not evoke different patterns of alpha power during left- and right-hand motor imagery during the first session, also are less likely to improve during the following training sessions. The investigation of whole-scalp differences allows a holistic comprehension of the neural basis of motor imagery. This method also characterizes a potential predictor of performance for future applications of MI-based neurofeedback and brain-computer interfaces.

Keywords: Human-Machine Interface, Human-Computer Interaction, User Interface Design
Abstract: Kinesthetic Motor Imagery (KMI) is a mental task which, if performed properly, can be very relevant in sports training or rehabilitation with a Brain-Computer Interface (BCI). Unfortunately, this mental task is generally complex to perform and can lead to a high degree of variability in its execution, reducing its potential benefits. The reason why the task of KMI is so difficult to perform is because there is no standardized way of instructing the subject in this mental task. This study presents an innovative BCI called Grasp-IT thought to support the learning of the KMI task, and the evaluation of two different learning methods: (i) a first one guided by an experimenter and based on the notion of progressiveness and (ii) a second one where the learners are alone and practice by trial and error. Our findings based on EEG analyses and subjective questionnaires validate the design of the Grasp-IT BCI and opens up perspectives on KMI learning modalities.

Keywords: Human Performance Modeling, Human-Machine Interface
Abstract: Motor Imagery-based Brain-Computer Interfaces(MI-BCI) allow users to control a computer for various applications using their brain activity alone, which is usually recorded by an electroencephalogram (EEG). Although BCI applications are numerous, their use outside laboratories is still scarce due to their poor accuracy. Some users cannot use BCIs, a phenomenon sometimes called “BCI illiteracy”, which impacts around 10% to30% of BCI users, who cannot produce discriminable EEG pat-terns. By performing neurophysiological analyses, and notably by identifying neurophysiological predictors of BCI performance, we may understand this phenomenon and its causes better. In turn, this may also help us to better understand and thus possibly improve, BCI user training. Therefore, this paper presents statistical models dedicated to the prediction of MI-BCI user performance, based on neurophysiological users’ features extracted from a two-minute EEG recording of a “relax with eyes open” condition. We consider data from 56 subjects that were recorded in a ‘relax with eyes open’ condition before performing a MI-BCI experiment. We used a machine learning regression algorithm with leave-one-subject-out cross-validation to build our model of prediction. We also computed different correlations between those features(neurophysiological predictors) and users’ MI-BCI performances. Our results suggest such models could predict user performances significantly better than chance (p≤0.01) but with a relatively high mean absolute error of 12.43%. We also found significant correlations between a few of our features and the performance, including the previously exploredμμμ-band predictor, as well as a new one proposed here: theμμμ-peak location variability. These results are thus encouraging to better understand and predict BCI illiteracy. However, they also require further improvements in order to obtain more reliable predictions.

Keywords: Human-Machine Interface
Abstract: Motor imagery (MI) classification is important as the emerging research interest of brain computer interface (BCI) due to its potential about real-world application. Advancing manipulation and control technology of external devices such as robotics, the need of MI for complex and human-like movements is growing. The two most important procedures that influence the performance of MI-BCI are feature extraction and classification. Although there have been recent studies on feature extraction for complex, there is no consensus on the classifier suitable for complex MI. This study aimed to identify the best classifier for complex MI decoding. Electroencephalography (EEG) recordings measured during complex MI, which are hand grasping, spreading, pronation and supination, were used for binary (grasp vs. twist) and quaternary classification. Time domain parameter, which have shown suitability for complex movement decoding in previous works, was used as the EEG feature. Four types of ten machine learning classifiers, which have been applied to MI-BCI, were compared. Shrinkage regularized linear discriminant analysis (SRLDA) exhibited the best classification accuracy in both binary (92.8%) and quaternary (55.2%). In the case of training and testing time, a small amount of time for real-time analysis were needed, except random forest and logistic regression. This study showed that SRLDA is an appropriate classifier for complex MI classification, due to its ability to handle stationary and high dimensionality feature, TDP. The findings suggest that complex MI-BCI could gain more benefit from applying linear and shrinkage regularized model (i.e., SRLDA).

Keywords: Human-Computer Interaction
Abstract: Electroencephalography (EEG) is a primary modality for estimating user intention in the brain-computer interface (BCI). In particular, BCI has been widely used to detect the intention of users in motor imagery (MI)-based tasks. Although the MI classification accuracy has been largely enhanced from previous efforts, MI-BCI studies have focused on extracting features only during MI tasks, not during the preparatory phases. The increment of alpha band power is induced by performing a task with attention. This study proposes a n approach for increasing MI-BCI performance by analyzing brain state in preparatory before the task. EEG recordings of nine healthy subjects from the open BCI dataset were investigated. The alpha lateralization index (ALI) was calculated for each trial and high ALI trials were utilized for learning lateralization-based model. MI classification accuracy using the lateralization-based model marked high performance (median accuracy = 63.2 %; interquartile range (IQR) = 50.0% - 54.8%) than the total trial-based approach (median accuracy = 52.0%; IQR = 50.0 % - 54.8%) with statistical significance (p = 0.018). This study suggests alpha lateralization which is an imbalance pattern between ipsilateral and contralateral is one of the main factors for improvement of performance. Accordingly, since the alpha liberalization before MI task could exert an effect on the MI phase, the analysis combined preparation with MI would derive highly benefit for the MI classification.

Keywords: Evolutionary Computation, Swarm Intelligence, Optimization
Abstract: Many real-world problems are categorized as large-scale problems, and metaheuristic algorithms as an alternative method to solve large-scale problem; they need the evaluation of many candidate solutions to tackle them prior to their convergence, which is not affordable for practical applications since the most of them are computationally expensive. In other words, these problems are not only large-scale but also computationally expensive, that makes them very difficult to solve. There is no efficient surrogate model to support large-scale expensive global optimization (LSEGO) problems. As a result, the algorithms should address LSEGO problems using a limited computational budget to be applicable in real-world applications. Coordinate Descent (CD) algorithm is an optimization strategy based on the decomposition of a n-dimensional problem into n one-dimensional problem. To the best our knowledge, there is no significant study to assess benchmark functions with various dimensions and landscape properties to investigate CD algorithm and compare with other metaheuristic algorithms. In this paper, we propose a modified Coordinate Descent algorithm (MCD) to tackle LSEGO problems with a limited computational budget. Our proposed algorithm benefits from two leading steps, namely, finding the region of interest and then shrinkage of the search space by folding it into the half with exponential speed. One of the main advantages of the proposed algorithm is being free of any control parameters, which makes it far from the intricacies of the tuning process. The proposed algorithm is compared with cooperative co-evolution with delta grouping on 20 benchmark functions with dimension 1000. Also, we conducted some experiments on CEC-2017, D=10, 30, 50, and 100, to investigate the behavior of MCD algorithm in lower dimensions. The results show that MCD is beneficial not only in large-scale problems, but also in low-scale optimization problems.

Keywords: Evolutionary Computation, Swarm Intelligence, Optimization
Abstract: Differential Evolution (DE) algorithm is an efficient metaheuristic algorithm in solving complex real-world optimization problems. DE algorithm benefits from two populations for updating individuals, while it might cause memory problems in practice during solving large-scale optimization problems; especially when they are used in an embedded system. One strategy to tackle this problem is utilizing a one-array scheme which benefits from only one population, leading to a half-space memory. This paper proposes a novel DE algorithm based on one-array DE and a random replacement strategy; it adds an additional competition to the selection operator to make better use of the new individual that it might be potentially noteworthy. The positive feature of the introduced replacement strategy is that it does not need any extra computational budget. Also, due to employing one-array strategy, the proposed scheme has a lower memory complexity. Our experiments on CEC-2017 benchmark function with dimensions 30, 50, and 100 clearly illustrate the effectiveness of the proposed DE algorithm.

Keywords: Evolutionary Computation, Swarm Intelligence, Optimization
Abstract: In society, the mutual influence and interaction between individuals constitute a social network, and opinion dynamics studies the generation, diffusion, and aggregation of ideas or behaviors in social networks. This paper introduces the ideas of evolution in opinion dynamics models into particle swarm optimization algorithm, and proposes a social opinion dynamics-guided particle swarm optimization algorithm (SODPSO). Firstly, in the process of population evolution, the idea of dynamic bounded confidence is used to select the learning object (the best individual in the confidence bound) for each particle to update, and for the individual whose learning object is itself, a difference operator is introduced to update it. Secondly, when the population stagnation reaches a certain threshold, the concepts of individual differences and acceptance are introduced. The particles are sorted and classified according to the fitness value, and different evolution strategies are used to update them in order to jump out of the current optimal solution. Finally, this paper compares SODPSO with the other five PSO variants on part of cec'17 benchmark functions. The experimental results demonstrate that the SODPSO proposed in this paper has greater advantages in functions with certain specific characteristics.

Keywords: Optimization, Evolutionary Computation, Heuristic Algorithms
Abstract: Electrical impedance tomography (EIT) is a noninvasive medical imaging procedure. Image reconstruction in EIT is difficult because it involves solving a non-linear and ill-posed mathematical problem. One of the most commonly implemented inverse approaches is usually a variation of the Newton Raphson algorithm. However, this approach is not guaranteed to reach a global optimum or a local optimum and as such, it requires an accurate initial estimation of the resistance distribution, which is not always available in practice.

In this paper, a new method is proposed to solve for the inverse problem in EIT while avoiding dependencies on the initial estimation of the resistance distribution. The proposed approach uses a differential evolution (DE) optimizer integrated with the Newton Raphson algorithm. The stochastic nature of DE allows the problem to be solved without having an accurate initial estimation and allows for solutions that will not be trapped in local minimal values. Simulation results indicate that the proposed approach outperforms the traditional differential evolution algorithm, and performs similarly to the traditional Modified Newton Raphson algorithm with accurate initial estimation. The proposed method does, however, have an advantage over the Modified Newton Raphson algorithm as it does not require an estimate of the initial resistance distribution.

Keywords: Swarm Intelligence, Evolutionary Computation, Computational Intelligence
Abstract: As a global optimization technique, the evolutionary algorithms provide powerful solvers to the data clustering problem. However, the individual representations in evolutionary clustering algorithms exist redundancy and inconsistency problems, which not only wastes the search efforts but also disorders the mutual learning process between individuals. To address the problems, in this paper, we propose a genetic sequence resorting method. This method first identifies a reference parameter vector and then resorts the encoding order of each individual according to their distance to the reference vector. In this way, the solutions represented by individuals are unified, which removes redundancy and enhances the efficiency of the mutual learning in the algorithm. Incorporating the above method, a new and generic evolutionary clustering framework is developed. Under this framework, we specifically design two algorithms: one for distance-based convex clustering and the other for density-based nonconvex clustering. Experiments show that our method can effectively improve the performance of evolutionary clustering algorithms on various datasets with both convex and nonconvex cluster shapes.

Keywords: Image Processing/Pattern Recognition, Machine Learning
Abstract: Motivated by its capacity to process complex characteristics and deal with nonlinear problems, tensor decompositions have been also introduced, recently, to treat remote sensing data. In this article, a new tensor formulation based feature extraction framework is suggested for hyperspectral target detection. The new proposed method includes the intrinsic decomposition, as tensor structures, to improve the hyperspectral data representation and get rid of the non-significant spatial proprieties. Besides of the joint exploitation of spectral and spatial content, the new proposed approach allows to extract more effective discriminative spatial features. A series of experiments, for the purpose of hyperspectral target detection, show that the suggested scheme can be conducted on hyperspectral images with satisfactory detection accuracies.

Keywords: Image Processing/Pattern Recognition, Machine Learning, Neural Networks and their Applications
Abstract: The bear attack to human beings is one of the fatal accidents, and it is becoming more critical to avoid such accidents because human's encountering a bear happens every year, even in a city area. It is required to discover bears quickly and warn people to avoid bear accidents. To realize sensor nodes that detect bears automatically using image recognition technology, we aim to realize an accurate and computationally-efficient bear-presence detection. In this paper, we propose a bear-presence detection method combining region proposal of a low-resolution and image classification. In the experiments, we show that the proposed method achieves 4.9% higher recall and 2.3% higher F-score than image classification without region-proposal. Moreover, the pro- posed method achieved 0.6% higher recall and 18.5% higher F-score than YOLOv3, which is one of state-of- the-art object detection methods while the execution time was reduced to 72.4% for bear images and 55.5% for non-bear images.

Keywords: Image Processing/Pattern Recognition, Machine Vision
Abstract: Point cloud registration a challenging task in situations with poor initial value and scenarios with limited geometric structure. In these cases, the correct correspondence between two point clouds is unknown and difficult to establish. To cope with this problem, the semantic of partial points is introduced in this paper. Firstly, the semantic information is used to find more reasonable correspondence, i.e. semantic point pairs. Secondly, we formulate a novel objective function to integrate the matching error of semantic point pairs as guidance of registration. Thirdly, a hyperparameter is applied to balance the confidence of semantic point pairs. At last, a novel algorithm under the ICP framework is presented to optimize the rigid transformation iteratively. The evaluation of KITTI data set reveals the robustness and accuracy of our method in the complex scenes mentioned above.

Keywords: Image Processing/Pattern Recognition, Machine Vision, Computational Intelligence
Abstract: Images captured in low light environments are more susceptible to loss of information in dark regions, making details of the scene not noticeable to humans. These dark images can also make difficult the use of automatic computer vision algorithms in applications as segmentation, object detection and recognition, or tracking. This paper proposes a Convolutional Neural Network (CNN) based method for the enhancement of weakly illuminated images. The network architecture estimates the illumination of the scene which is further used to enhance the images using the Retinex model. The experiments conducted in datasets with synthetic and natural images proved that our method surpassed other state of the art approaches (quantitatively and qualitatively), creating images with less color distortion.

Keywords: Image Processing/Pattern Recognition, Machine Vision, Neural Networks and their Applications
Abstract: Automated herb identification plays an important role in protecting and investigating the herbs for botanists, which has been widely applied in the field of cosmetic, medical and food industry areas. Traditionally, due to the complicated background and various herb patterns, herb discriminative feature extractions is a hard work. And some existing methods may not be applicable when the herbs exist in actual wild environment. Therefore, how to locate the valid herb regions and extract the effective features is an open issue. In this paper, a novel CNN model is proposed for herb identification by using the part-information perception module (PPM) and species classification module (SCM). A new attention mechanism, namely part priority attention mechanism (PPAM), is proposed by training PPM independently with herb part labels. It should be pointed out that the proposed PPAM can guide the model to focus on the position of herb parts and suppress the irrelevant noisy regions. Moreover, depthwise separable convolution and label smoothing technique are introduced to decrease the model complexity and regularize the impact of mistake labels. In the experiment part, a large-scale herb dataset is constructed, which consists many kinds of challenging herb species in the wild environment. Additionally, these images contains both species labels and part labels. Experimental results demonstrate that our model achieves obviously improvement in term of accuracy and model size compared with other classical deep learning models.

Keywords: Knowledge Acquisition in Intelligent
Abstract: Various relations exist in the recent big data and are important in understanding enormous information. We have proposed an analytic framework of relations among unstructured data focusing on miss-classified items in clustering algorithms. The framework has succeeded in finding out somewhat relations among substances. The analytical framework, however, could only perform the subjective analysis but quantitative analysis of those relations.In this paper, we, therefore, have developed the quantitative analytic framework of those relations. We also carried out experiments using news articles and quantitative analysis of the relations among nations following the previous work.

Keywords: Knowledge Acquisition in Intelligent, Neural Networks and their Applications, Computational Intelligence
Abstract: Accurate and efficient measurements for evaluating the similarity between mathematical formulae play an important role in mathematical information retrieval. Most previous studies have focused on representing formulae in different types to catch their features and combining the traditional structure matching algorithms. This paper presents a new unsupervised model called N-ary Tree-based Formula Embedding Model (NTFEM) for the task of mathematical similarity evaluation. Using an n-ary tree structure to represent the formula, we convert the formula into a linear sequence that can be viewed as the input sentence and then embed the formula by using a word embedding model. Based on the characteristics of mathematical formulae, a weighting function is also used to get the final weighted average embedding vector. Through some experiments on NTCIR-12 Wikipedia Formula Browsing Task, our model can outperform previous formula search engines in Bpref prediction metrics. In addition, compared with traditional tree-based models, our model not only improves the retrieval effect but also greatly reduces the training time and improves training efficiency.

Keywords: Swarm Intelligence, Neural Networks and their Applications, Computational Intelligence
Abstract: The purpose of this paper is to compare the abilities and deficiencies of various swarm based algorithms for training artificial neural networks. This paper uses seven algorithms, seven regression problems, sixteen classification problems, and four bounded activation functions to compare algorithms in regards to loss, accuracy, hidden unit saturation, and overfitting. It was found that particle swarm optimization is the top algorithm for regression problems based on loss, firefly algorithm was the top algorithm for classification problems when examining accuracy and loss. The ant colony optimization and artificial bee colony algorithms caused the least amount of hidden unit saturation, with the bacterial foraging optimization algorithm producing the least amount of overfitting.

Keywords: Intelligent Internet Systems, Computational Intelligence, Machine Learning
Abstract: Adaptive recommender systems are increasingly showing their importance as profiling is a dynamic problem. Their goal is to update recommendation models as new interactions take place, thus swiftly adapting to drifts in the user’s behavior and desires, and item’s audience. However, existing recommendation algorithms usually do not perform well during drifts, as they take long to adapt to changes, or these updates are suboptimal since they account for all profiles’ preferences equally, which is often untrue as each individual and its changes are unique. In this paper, we propose the ADADRIFT algorithm to deal with user and item-based drifts in adaptive recommender systems using personalized learning rates based on profile statistics. The experiments using stream-based recommender systems (ISGD and BRISMF) across four different datasets show that ADADRIFT surpasses ADADELTA with significant improvements in recommendation rates. The best results appear when the data streams have a long history of the users’ or items’ interactions and drifts become noticeable. The experimentation in this work highlight the importance of handling drifts in recommender systems.

Keywords: Machine Learning, Computational Intelligence, Cybernetics for Informatics
Abstract: Detection of cyber attacks in the Internet of Things (IoT) networks has lately been a growing concern. Due to the extensive use of IoT infrastructures in numerous domains, these malicious attacks are also increasing continuously and changing over time. Moreover, devices connected in IoT networks are operated without any human intervention for longer times. Hence, intelligent network-based security solutions are very important to provide timely detection of these attacks to protect an IoT system from potential failure. Different machine learning based techniques have already been proposed to provide effective solution to discover and counteract network intrusion aiming to ensure security in the network. In the context of IoT networks, little attention has been paid to the identification of malicious attacks. To this end, we propose an effective intrusion detection system (IDS) to detect unforeseen IoT cyberattacks by using various bagging and boosting ensemble methods and feed forward artificial neural network. We have used a recently published dataset, UNSW-NB15, containing simulated IoT sensor data to estimate the performance of the proposed models through 5-fold cross validation technique. The performance results show the effectiveness of the models with a small set of automatically selected optimal features from the dataset.

Keywords: Machine Learning, Neural Networks and their Applications, Evolutionary Computation
Abstract: As one of the most important renewable energy sources in the world, wind energy has been widely studied and applied. When using wind energy to generate electricity, it will be of great help to the safety and stability of power supply, if the wind speed in the future can be accurately known. In this paper, a new short-term wind speed prediction method based on historical data is proposed. Firstly, the wind speed is pre-processed through variational mode decomposition (VMD) of which the parameters are optimized using a multi-objective optimization method in this paper, owing to the effect of VMD is greatly affected by parameters. Then, the combined forecasting method combining support vector machine improved by particle swarm optimization algorithm (PSO-SVM), back propagation neural network (BP) and long short-term memory network (LSTM) is used to predict each wind speed component. This paper takes data sets available at the US Virgin Islands Bovoni measurement station as an example, conducting experiments and performing analysis. Compared with other prediction models, it is demonstrated that the model proposed in this paper significantly improves the prediction accuracy. Also, the wind speed in January, April, July and October are forecasted respectively to test the stability of models, and the result shows that the proposed model has the best adaptability.

Keywords: Machine Learning, Neural Networks and their Applications, Image Processing/Pattern Recognition
Abstract: Our goal is to estimate athlete 3D pose from monoc-ular TV sports video at a lower computational cost. To achievethis goal, we utilize a pre-trained deep neural network as a 3Dpose estimator for estimating human 3D pose from 2D jointlocations of the person in each image. Each image in populardatasets used for training such 3D pose estimator is obtainedfrom a camera whose axis is parallel to the ground. On the otherhand, since an image in TV sports video is generally taken froma bird’s eye view, joint locations of a human is distorted in thelower part of the image. Therefore, it is not appropriate to give2D joint locations of the person directly to the pre-trained 3Dpose estimator. To resolve this problem, we propose to correct 2Djoint locations in an image of TV sports video by a homographytransformation that maps the points in the image of TV sportsvideo to the corresponding points in the image taken by thecamera that captures training data for the 3D pose estimator.Experimental results show that the proposed method can estimateathlete 3D pose from monocular TV sports video.

Keywords: Machine Learning, Neural Networks and their Applications, Industry 4.0
Abstract: It is common for a time series dataset to have missing values, and it is necessary to fill these missing elements before fitting any model for forecasting or prediction. Time series imputation remains a challenging task due to the existence of non-linear dependencies between current and past values. Conventional methods, such as deletion of rows containing missing values or filling them with the last observed value, add bias to the data and are therefore inefficient. There are situations where data is missing at consecutive points or random points in the dataset, and one particular method may not work well for all cases. In this paper, nine commonly used models in the field of imputation, based on tools of statistics, machine learning, and deep learning, are compared. Results show that Linear Memory Vector Gated Recurrent Unit (LIME-GRU) outperforms the other tested models by having the least Mean Square Error (MSE) and Root Mean Squared Error (RMSE). A predictive model to gauge the impact of imputation on prediction is also used to validate the findings. The results of the prediction model illustrate that with LIME-GRU, there was a 39% improvement in Average Aggregated Measure (AAGM) when compared with mode imputation on a particular test case.

Keywords: Machine Learning, Neural Networks and their Applications, Industry 4.0
Abstract: Water distribution networks have shown an increased rate of failure due to material deterioration. In this paper, we apply a Recurrent Neural Hawkes Process model to learn the failure intensity function of water pipes. The failure intensity function is learned based on two components: the base failure rate that is determined by the unique pipe profile attributes, and the effect of past failures. Compared to the existing solutions, our model is able to predict the time to next failure on an individual water pipe level. The learned failure intensity function is used to identify value points in the deterioration process of water pipes that represent their economical end-of-life. We use data from a Dutch water distribution network that consists of 49,600 km of pipelines to test the performance of the proposed model. We have made this dataset available online.

Keywords: Machine Learning, Neural Networks and their Applications, Machine Vision
Abstract: We propose a new framework that boasts of low training costs and high generalization performance in capturing human action expressions simultaneously on spatial and temporal structures. First, a video slicing process has to be established. Then, in order to capture the divergence and likelihood expression of the spatial structure in each video slice, a pipeline is introduced using a pre-trained CNN. In addition, any pre-trained network can be used to extract these features. Subsequently, Linear Dynamical Systems (LDS) is established to determine the timing relationship between two adjacent slices to obtain the temporal structure of the divergence and likelihood features, which are expressed as LD-Divergence and LD-Likelihood, respectively. In the UCF50 and UCF101 datasets, we analyzed the impact of different feature dimensions retained by PCA on recognition. Finally, we combined LD-Divergence and LD-Likelihood to improve accuracy to 0.961 and 0.949 on the UCF50 and UCF101 datasets, respectively. Experimental results show that the proposed framework simultaneously expresses spatial and temporal structures, which in turn produce state-of-the-art results.

Keywords: Neural Networks and their Applications, Optimization, Industry 4.0
Abstract: One of the main objectives of production planning is to minimize the usage of resources and manufacturing-related costs while meeting the customer's requirements, such as delivery dates and quality. Production planners deal with various scheduling problems that are often NP-hard and can not be optimally solved by humans. Solving such problems often relies on methods from the Operations Research (OR) field. Recently, Neural Combinatorial Optimization (NCO) has emerged as a promising field of research that aims at tackling different optimization tasks using the latest advancements in machine learning, including deep reinforcement learning. These methods can be successfully used for short-term production planning because of their flexibility and speed. In this paper, we examine the applicability and scalability of neural combinatorial optimization methods in the context of production planning. We define an evaluation metric to investigate the stability and quality of the solutions. Furthermore, we develop an experimental setup allowing to compare various approaches for production scheduling with sequence-dependent setup costs under real-world production conditions. Although an optimality gap is observed when compared to established OR methods, our experiments demonstrate the superiority of NCO in terms of scheduling time.

Keywords: Fuzzy Systems and their applications, Machine Learning, Neural Networks and their Applications
Abstract: Prediction interval (PI) as a method of probabilistic prediction can output the prediction range with a certain degree of confidence. It can give the users more information than point prediction. The noise of data in PI is usually assumed as a Gaussian, Laplace or other single distribution. However, these assumptions are not suitable for all the applications. In order to solve this problem, a mixed approach based on Long Short Term Memory Network with bootstrap (LSTM-bootstrapping) and mixed Gaussian distribution (MGD) with Expectation-Maximization (EM) algorithm is proposed to forecast intervals for time series. LSTM is chosen here because of its extremely effectiveness for time series prediction. Firstly, LSTM-bootstrapping is employed to calculate the model uncertainties and the point prediction. Afterwards, we assume that the noise satisfies a mixed Gaussian distribution and the EM algorithm is applied to estimate the noise uncertainty. Then PI can be acquired by the variances of model and noise uncertainty. The proposed predictive approach is evaluated on wind speed, heteroscedastic wind power and reg capacity price datasets. The results show that our method can solve the uncertainty problem of arbitrary distribution and obtain better performance.

Keywords: Neural Networks and their Applications, Optimization
Abstract: The embedding generated by pre-trained models has attracted the attention of many scholars in the past few years. Most of the context-sensitive embeddings have confirmed the positive impact on some basic tasks of classification, which have only a few types. In this paper, we make an empirical comparison of different pre-trained embeddings on the task of ultra-fine entity typing which has more than 10k types. We apply 7 kinds of pre-trained embedding to the typing model to prove whether the pre-trained embedding has a positive effect. The results indicate that almost all context-sensitive embeddings improve the performance of models using Glove. The pre-trained embedding generated by BERT achieves the best performance in the Ultra-Fine dataset and OntoNotes dataset, which shows BERT has better capability to extract finer-grained information than other pre-trained models.

Keywords: Neural Networks and their Applications, Optimization, Machine Learning
Abstract: Recurrent neural networks that can capture temporal characteristics on multiple timescales are a key architecture in machine learning solutions as well as in neurocognitive models. A crucial open question is how these architectures can adopt both multi-term dependencies and systematic fluctuations from the data or from sensory input, similar to the adaptation and abstraction capabilities of the human brain. In this paper, we propose an extension of the classic Continuous Time Recurrent Neural Network (CTRNN) by allowing it to learn to gate its timescale characteristic during activation and thus dynamically change the timescales in processing sequences. This mechanism is simple but bio-plausible as it is motivated by the modulation of oscillation modes between neural populations. We test how the novel Gating Adaptive CTRNNs can solve difficult synthetic sequence prediction problems and explore the development of the timescale characteristics as well as the interplay of multiple timescales. As a particularly interesting finding, we report that timescale distributions emerge, which simultaneously capture systematic patterns as well as spontaneous fluctuations. Our extended architecture is interesting for cognitive models that aim to investigate the development of specific timescale characteristic under temporally complex perception and action, and vice versa.

Keywords: Biometric Systems and Bioinformatics, Cybernetics for Informatics
Abstract: Essential tremor (ET) is one of the most common movement disorders. Undesired involuntary periodic movements are generated in ET patients. Some devices have been developed to suppress ET movements up to the present. In this paper, Tonic Vibration Reflex (TVR) is studied to investigate the possibility to suppress ET movements. The TVR is a phenomenon of human reflex which is induced by applying mechanical vibration stimulation to a human muscle. If a counter-phase movement of the ET movement of the patient is artificially generated by the TVR, the ET movement would be canceled. In order to investigate its possibility, fundamental study on the TVR is carried out in this study. In order to confirm the possibility of generating the TVR in forearm pronation/supination, the effect of the amplitude and the range of frequency of vibration stimulation on the TVR in forearm pronation/supination is studied in this paper. The effect of forearm posture is also studied. Furthermore, the effect of the on-and-off vibration stimulation on the TVR in forearm pronation/supination is investigated. The experimental results show the possibility of tremor movement suppression with the TVR.

Keywords: Information Assurance & Intelligent, Computational Life Science, Computational Intelligence
Abstract: Blind people use screen readers (SRs) to manipulate their personal computers. Due to the various voices and sounds handled by the screen reader, the more application windows open, the heavier the cognitive burden it feels, and the lower its performance becomes. In this paper, we propose a support scheme that enables SR users with blindness to effectively control multiple windows. As a proof-of-concept system, we propose a set of shortcut key buttons that provide direct jumps to the targeted window through a simple action. According to the evaluation results, our shortcut key buttons had positive effects on reducing the subjective workload and the number of keystrokes.