Summary Nonlinear systems widely exist in real‐word applications and the research for these has enjoyed a long fruitful history, including system identification community. However, modeling nonlinear is often quite challenging still remains many unresolved questions. This article considers online issue of Hammerstein systems, whose static function modeled by B‐spline network. First, model studied constructed using bilinear parameter decomposition model. Second, recursive algorithms are...

Abstract The nonlinear autoregressive exogenous (NARX) model shows a strong expression capacity for systems since these have limited information about their structures. However, it is difficult to the NARX system with relatively high dimension by using popular polynomial and neural network efficiently. This article uses tensor B‐spline (TNBS) system, whose representation of multivariate weight can alleviate computation store burden processing high‐dimensional systems. Furthermore, applying...

To facilitate human gait recognition, this paper proposes a new frontal-view recognition method using dynamics and deep learning. Rather than adopting lateral-view parameters as features in the literature, we employ improved classification methods to avoid rate drops due complicated surveillance environment. Specifically, characterize binary walking silhouettes with three different kinds of features, including kinematic spatial ratio area features. In addition, capture underlying...

Abstract Despite that much progress has been reported in gait recognition, most of these existing works adopt lateral-view parameters as features, which requires large area data collection environment and limits the applications recognition real-world practice. In this paper, we frontal-view walking sequences rather than propose a new method based on multi-modal feature representations learning. Specifically, characterize with two different kinds features representations, including holistic...

In this paper, a novel approach, which is based on an improved state refinement for long short-term memory(LSTM) determined 3D convolution-attention model, proposed myocardial infarction detection. Theproposed LSTM CAISR-LSTM model trained in end-to-endfashion. The input 12-lead electrocardiogram (ECG) signals are preprocessed to remove high-frequency noiseand baseline drift. Then, the ECG transformed into time-frequency images using continuous wavelettransform (CWT) and bilinear...

Abstract Risk stratification of hypertension plays a crucial role in the treatment decisions
and medication guidance during clinical practices. Although fruitful achievements have been
reported on risk hypertension, potential use ambulatory blood pressure
monitoring data is not well investigated. Different from single measuring pressure
data, long-term pressure monitoring can provide more comprehensive dynamical
blood information. Therefore, this paper...

Gait is an important biometric technology for human identification at a distance. This study focuses on gait features obtained by Microsoft Kinect and proposes new model-based recognition method combining deterministic learning theory the data stream of Kinect. Deterministic employed to capture dynamics underlying Kinect-based parameters. Spatial-temporal can be represented as trajectories spatial-temporal parameters, which implicitly reflect temporal changes silhouette shape. Kinematic...

Epilepsy, as a sudden and life-threatening nervous system disease, seriously affected around 6% population in the world. Epileptic classification has attracted wide attention past number of methods have been developed. But currently studies are mainly on three epileptic states (preictal, ictal, interictal) or seizure/non-seizure detection. Among them, one hour before seizure onset was generally considered preictal, where division is actually not fine enough for some practical applications....

This paper presents a new gait recognition method based on acceleration and angular velocity data captured by inertial sensors via deterministic learning. These features describe the motion trajectories of human contain rich information for persons identification. The approach consists two phases: training phase phase. In phase, dynamics underlying different individuals' gaits are represented features, locally accurately approximated radial basis function (RBF) neural networks. obtained...

Gait recognition plays an important role in the area of biometric recognition. Despite that much progress has been made for gait recent years, most them are based on lateral-view characteristics. These methods usually require a large data collection to capture full sequences, which only applicable wide outdoor spaces. In this paper, we propose new frontal-view method human dynamics and deep transfer learning. The frontal silhouettes characterized with four kinds time-varying features, namely...

This paper considers the mean-square leader-following consensus problem for heterogeneous multi-agent systems (MASs) affected by noises. Because only output signal of a leader can be measured, distributed dynamic control driven an event-triggered mechanism (ETM) is proposed. The proposed ETM reduce transmission amount from sampler to controller. Using tine-delay model method, system under controller modelled as stochastic time-delay system. Then time-varying Hanaly inequality utilised obtain...

Model construction, fitting and feature extraction remain the main difficult tasks in model-based gait recognition methods. To develop an easy practical system, this paper investigate combined use of deterministic learning Microsoft Kinect sensor for human recognition. First, real-time data stream joints position are acquired by using MATLAB toolbox. Second, four different time-varying selected to represent walking. Third, variability underlying individuals' features is effectively modeled...

This paper proposes an effective and rapid human gait recognition system based on deterministic learning Microsoft sensor. In order to deal with the difficulties of feature extraction in system, Kinect sensor is used for realtime skeleton detection tracking. Dynamical algorithm implementation achieved a scheme proposed. The graphical programming language C# MATLAB GUI controls are combined building user-friendly simple interfaces display training process. results detailed parameters can be...

Deformation of gait silhouettes caused by different view angles heavily affects the performance recognition. In this paper, a new method based on deterministic learning and knowledge fusion is proposed to eliminate effect angle for efficient view-invariant First, binarized walking are characterized with three kinds time-varying width parameters. The nonlinear dynamics underlying individuals' parameters effectively approximated radial basis function (RBF) neural networks through algorithm....

A frontal-view gait recognition approach based on Kinect features and deterministic learning is presented in this paper. The utility of sensor eliminates the adverse interference background for feature extraction. Different kinds are extracted dynamics underlying different individual's time-varying captured by using radial basis function (RBF) neural networks (NNs) through (DL). obtained knowledge system stored constant RBF networks. bank estimators constructed to represent training...