Self-learning sliding mode control based on adaptive dynamic programming for nonholonomic mobile robots.

This paper proposes a self-learning sliding mode control (SlSMC) strategy with stability guarantee for the trajectory tracking of nonholonomic mobile robots (NMRs) under matched uncertainties, which improves the control performance of NMRs by optimizing the reaching law and the sliding mode surface of SMC as well as retaining the finite-time convergence and the robustness to uncertainties. In the presence of adverse factors such as skidding, slipping and environmental noise, the kinematic model of NMRs is reconstructed and an integral terminal sliding mode controller is designed for the trajectory tracking of NMRs. Then, based on the sliding mode controller, the proposed control strategy formulates the optimization of the SMC's reaching law and the sliding mode surface under stability constraints as two asynchronous optimal control problems with control constraints. Meanwhile, an online continuous-time receding-horizon optimization mechanism based on an actor-critic algorithm is proposed to solve the optimal problems asynchronously and improve online learning efficiency. The stability and the convergence of the proposed strategy are validated both in theory and simulations. Furthermore, extensive contrastive simulation results illustrate that the proposed receding horizon learning-based control strategy outperforms three recent methods in control performance. Finally, experiments of the proposed self-learning SMC strategy are carried out based on a real intelligent vehicle, and the experimental results also verify that the proposed method can meet the actual control needs of NMRs.

MRCG: A MRI Retrieval Framework With Convolutional and Graph Neural Networks for Secure and Private IoMT.

In the context of Industry 4.0, the medical industry is horizontally integrating the medical resources of the entire industry through the Internet of Things (IoT) and digital interconnection technologies. Speeding up the establishment of the public retrieval database of diagnosis-related historical data is a common call for the entire industry. Among them, the Magnetic Resonance Imaging (MRI) retrieval system, which is one of the key tools for secure and private the Internet of Medical Things (IoMT), is significant for patients to check their conditions and doctors to make clinical diagnoses securely and privately. Hence, this paper proposes a framework named MRCG that integrates Convolutional Neural Network (CNN) and Graph Neural Network (GNN) by incorporating the relationship between multiple gallery images in the graph structure. First, we adopt a Vgg16-based triplet network jointly trained for similarity learning and classification task. Next, a graph is constructed from the extracted features of triplet CNN where each node feature encodes a query-gallery image pair. The edge weight between nodes represents the similarity between two gallery images. Finally, a GNN with skip connections is adopted to learn on the constructed graph and predict the similarity score of each query-gallery image pair. Besides, Focal loss is also adopted while training GNN to tackle the class imbalance of the nodes. Experimental results on some benchmark datasets, including the CE-MRI dataset and a public MRI dataset from the Kaggle platform, show that the proposed MRCG can achieve 88.64% mAP and 86.59% mAP, respectively. Compared with some other state-of-the-art models, the MRCG can also outperform all the baseline models.

Semi-Supervised 3D Medical Image Segmentation Based on Dual-Task Consistent Joint Learning and Task-Level Regularization.

Semi-supervised learning has attracted wide attention from many researchers since its ability to utilize a few data with labels and relatively more data without labels to learn information. Some existing semi-supervised methods for medical image segmentation enforce the regularization of training by implicitly perturbing data or networks to perform the consistency. Most consistency regularization methods focus on data level or network structure level, and rarely of them focus on the task level. It may not directly lead to an improvement in task accuracy. To overcome the problem, this work proposes a semi-supervised dual-task consistent joint learning framework with task-level regularization for 3D medical image segmentation. Two branches are utilized to simultaneously predict the segmented and signed distance maps, and they can learn useful information from each other by constructing a consistency loss function between the two tasks. The segmentation branch learns rich information from both labeled and unlabeled data to strengthen the constraints on the geometric structure of the target. Experimental results on two benchmark datasets show that the proposed method can achieve better performance compared with other state-of-the-art works. It illustrates our method improves segmentation performance by utilizing unlabeled data and consistent regularization.

Self-Paced Dynamic Infinite Mixture Model for Fatigue Evaluation of Pilots' Brains.

Current brain cognitive models are insufficient in handling outliers and dynamics of electroencephalogram (EEG) signals. This article presents a novel self-paced dynamic infinite mixture model to infer the dynamics of EEG fatigue signals. The instantaneous spectrum features provided by ensemble wavelet transform and Hilbert transform are extracted to form four fatigue indicators. The covariance of log likelihood of the complete data is proposed to accurately identify similar components and dynamics of the developed mixture model. Compared with its seven peers, the proposed model shows better performance in automatically identifying a pilot's brain workload.

Fatigue Detection of Pilots' Brain Through Brains Cognitive Map and Multilayer Latent Incremental Learning Model.

This work proposes a nonparametric prior induced deep sum-logarithmic-multinomial mixture (DSLMM) model to detect pilots' cognitive states through the developed brain power map. DSLMM uses multinormal distribution to infer the latent variable of each neuron in the first layer of the network. These latent variables obeyed a sum-logarithmic distribution that is backpropagated to its observation vector and the number of neurons in the next layer. Multinormal distribution is used to segment the extended observation vector to form a matrix associated with the width of the next layer. This work also proposes an adaptive topic-layer stochastic gradient Riemann (ATL-SGR) Markov chain Monte Carlo (MCMC) inference method to learn its global parameters without heuristic assumptions. The experimental results indicate that DSLMM can extract more probability distribution contained in the brain power map layer by layer, and achieve higher pilot cognition detection accuracy.

Scalable Gamma-Driven Multilayer Network for Brain Workload Detection Through Functional Near-Infrared Spectroscopy.

This work proposes a scalable gamma non-negative matrix network (SGNMN), which uses a Poisson randomized Gamma factor analysis to obtain the neurons of the first layer of a network. These neurons obey Gamma distribution whose shape parameter infers the neurons of the next layer of the network and their related weights. Upsampling the connection weights follows a Dirichlet distribution. Downsampling hidden units obey Gamma distribution. This work performs up-down sampling on each layer to learn the parameters of SGNMN. Experimental results indicate that the width and depth of SGNMN are closely related, and a reasonable network structure for accurately detecting brain fatigue through functional near-infrared spectroscopy can be obtained by considering network width, depth, and parameters.

Rotated Sphere Haar Wavelet and Deep Contractive Auto-Encoder Network With Fuzzy Gaussian SVM for Pilot's Pupil Center Detection.

How to track the attention of the pilot is a huge challenge. We are able to capture the pupil status of the pilot and analyze their anomalies and judge the attention of the pilot. This paper proposes a new approach to solve this problem through the integration of spherical Haar wavelet transform and deep learning methods. First, considering the application limitations of Haar wavelet and other wavelets in spherical signal decomposition and reconstruction, a feature learning method based on the spherical Haar wavelet is proposed. In order to obtain the salient features of the spherical signal, a rotating spherical Haar wavelet is also proposed, which has a consistent scale in the same direction between the reconstructed image and the original image. Second, in order to find a better characteristic representation of the spherical signal, a higher contractive autoencoder (HCAE) is designed for the potential representation of the spherical Haar wavelet coefficients, which has two penalty items, respectively, from Jacobian and two order items from Taylor expansion of the point x for the contract learning of sample space. Third, in order to improve the classification performance, this paper proposes a fuzzy Gaussian support vector machine (FGSVM) as the top classification tool of the deep learning model, which can punish some Gaussian noise from the output of the deep HCAE network (DHCAEN). Finally, a DHCAEN-FGSVM classifier is proposed to identify the location of the pupil center. The experimental results of the public data set and actual data show that our model is an effective method for spherical signal detection.

Nonparametric Bayesian Prior Inducing Deep Network for Automatic Detection of Cognitive Status.

Pilots' brain fatigue status recognition faces two important issues. They are how to extract brain cognitive features and how to identify these fatigue characteristics. In this article, a gamma deep belief network is proposed to extract multilayer deep representations of high-dimensional cognitive data. The Dirichlet distributed connection weight vector is upsampled layer by layer in each iteration, and then the hidden units of the gamma distribution are downsampled. An effective upper and lower Gibbs sampler is formed to realize the automatic reasoning of the network structure. In order to extract the 3-D instantaneous time-frequency distribution spectrum of electroencephalogram (EEG) signals and avoid signal modal aliasing, this article also proposes a smoothed pseudo affine Wigner-Ville distribution method. Finally, experimental results show that our model achieves satisfactory results in terms of both recognition accuracy and stability.

Identification of Autistic Risk Candidate Genes and Toxic Chemicals via Multilabel Learning.

As a group of complex neurodevelopmental disorders, autism spectrum disorder (ASD) has been reported to have a high overall prevalence, showing an unprecedented spurt since 2000. Due to the unclear pathomechanism of ASD, it is challenging to diagnose individuals with ASD merely based on clinical observations. Without additional support of biochemical markers, the difficulty of diagnosis could impact therapeutic decisions and, therefore, lead to delayed treatments. Recently, accumulating evidence have shown that both genetic abnormalities and chemical toxicants play important roles in the onset of ASD. In this work, a new multilabel classification (MLC) model is proposed to identify the autistic risk genes and toxic chemicals on a large-scale data set. We first construct the feature matrices and partially labeled networks for autistic risk genes and toxic chemicals from multiple heterogeneous biological databases. Based on both global and local measure metrics, the simulation experiments demonstrate that the proposed model achieves superior classification performance in comparison with the other state-of-the-art MLC methods. Through manual validation with existing studies, 60% and 50% out of the top-20 predicted risk genes are confirmed to have associations with ASD and autistic disorder, respectively. To the best of our knowledge, this is the first computational tool to identify ASD-related risk genes and toxic chemicals, which could lead to better therapeutic decisions of ASD.

DMMAN: A two-stage audio-visual fusion framework for sound separation and event localization.

Videos are used widely as the media platforms for human beings to touch the physical change of the world. However, we always receive the mixed sound from the multiple sound objects, and cannot distinguish and localize the sounds as the separate entities in videos. In order to solve this problem, a model named the Deep Multi-Modal Attention Network (DMMAN), is established to model the unconstrained video datasets for further finishing the sound source separation and event localization tasks in this paper. Based on the multi-modal separator and multi-modal matching classifier module, our model focuses on the sound separation and modal synchronization problems using two stage fusion of the sound and visual features. To link the multi-modal separator and multi-modal matching classifier modules, the regression and classification losses are employed to build the loss function of the DMMAN. The estimated spectrum masks and attention synchronization scores calculated by the DMMAN can be easily generalized to the sound source and event localization tasks. The quantitative experimental results show the DMMAN not only separates the high quality of the sound sources evaluated by Signal-to-Distortion Ratio and Signal-to-Interference Ratio metrics, but also is suitable for the mixed sound scenes that are never heard jointly. Meanwhile, DMMAN achieves better classification accuracy than other contrast baselines for the event localization tasks.