检索结果-内蒙古大学图书馆

A Multi-Kernel-Based Multi-View Deep Non-Negative Matrix Factorization for Enhanced Healthcare Data Clustering

学校读者我要写书评

暂无评论

IEEE Transactions on Consumer Electronics 2024年 1-1页

作者： Che, Hangjun Yang, Xuanhao College of Electronic and Information Engineering Southwest University and Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing Chongqing China College of Electronic and Information Engineering Southwest University Chongqing China

Multi-view clustering methods based on deep matrix factorization play a vital role in data analysis within the healthcare sector. However, existing methods predominantly conduct deep matrix factorization in the original data space, which is not conducive to addressing non-linear and complex data patterns. To address this issue, the Multi-kernel based Multi-view Deep Non-negative Matrix Factorization with Optimal Consensus Graph (OGMKMDNMF) is introduced. This approach utilizes deep non-negative matrix factorization after projecting the data matrix into a high-dimensional kernel space. Additionally, it employs optimal consensus graph to alleviate the detrimental effects arising from misassigned nearest neighbors during the construction of similarity matrix. An innovative iterative optimization algorithm is developed for OGMKMDNMF. The experimental results demonstrate the effectiveness and competitive advantage of OGMKMDNMF in addressing multi-view healthcare data clustering tasks. IEEE

关键词： Cluster analysis

Stress Recognition Based on the Triple-axis Acceleration

学校读者我要写书评

暂无评论

Stress Recognition Based on the Triple-axis Acceleration

2016 IEEE International Conference on Progress in Informatics and Computing

作者： Hong Liu Wanhui Wen Guangyuan Liu Jie Zhang Jialan Xie School of Electronic and Information Engineering Southwest University Chongqing Key Laboratory of Nonlinear Ciruits and Intelligent Processing

Strong and weak stress were elicited by real-scene thesis defense and pre-defending presentation of the thesis work, and tri-axial accelerometer data were acquired and analyzed for their ability to recognize strong and weak stress. Twenty-six subjects(7 females and 19 males) participated in the data acquisition experiment. A support vector machine classifier obtained a correct rate of 92.31% in the binary classification of the 26 subjects' tri-axial accelerometer data acquired during strong stress and weak stress.

关键词： emotion recognition tri-axial acceleration stress support vector machine

Mean Square Exponential Stabilization Analysis of Stochastic Neural Networks with Saturated Impulsive Input

学校读者我要写书评

暂无评论

SSRN 2023年

作者： Deng, Hao Li, Chuandong Chang, Fei Wang, Yinuo Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Southwest University Chongqing400715 China

The exponential stabilization of stochastic neural networks in mean square sense with saturated impulsive input is investigated in this paper. Firstly, the saturated term is handled by polyhedral representation method. When the impulsive sequence is determined by average impulsive interval, impulsive density and mode-dependent impulsive density, the sufficient conditions for stability are proposed, respectively. Then, the ellipsoid or the polyhedron are used to estimate the attractive domain, respectively. By transforming the estimation of the attractive domain into a convex optimization problem, a relatively optimum domain of attraction is obtained. Finally, a three-dimensional continuous time Hopfield neural network example is provided to illustrate the effectiveness and rationality of our proposed theoretical results. © 2023, The Authors. All rights reserved.

关键词： Hopfield neural networks

Synchronization of Reaction-Diffusion Neural Networks with Sampled-Data Control Via a New Two-Sided Looped Functional

学校读者我要写书评

暂无评论

SSRN 2022年

作者： Huan, Mingchen Li, Chuandong Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Southwest University Chongqing400715 China

In this paper, the exponential synchronization problem of reaction-diffusion neural networks with sampled-data control is addressed via looped-functional approach. Considering the transmission delay of sampled-data controller, a two-sided looped functional is designed to obtain thesynchronization conditions, which are less conservative than those with the traditional Lyapunov method. The research results are applied to Takagi-Sugeno (T-S) fuzzy models with reactiondiffusion terms. Three numerical examples are presented to showthe feasibility and effectivenessof our methods by using looped functional. © 2022, The Authors. All rights reserved.

关键词： Lyapunov methods

Accelerated Row-stochastic Optimization over Directed Graphs with Uncoordinated Step Sizes

学校读者我要写书评

暂无评论

Accelerated Row-stochastic Optimization over Directed Graphs...

International Conference on Control and Automation (ICCA)

作者： Kang Zhou Zhenyuan Du Huaqing Li Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Southwest University Chongqing PR China

ISBN: (数字)9781728190938

ISBN: (纸本)9781728190945

This paper investigates a distributed optimization problem over a multiagent network, in which the target of agents is to collaboratively optimize the sum of all local objective functions. The case discusses that the network topology among agents is described by a strongly connected directed graph. The proposed algorithm utilizes row-stochastic weight matrices and uncoordinated step sizes. Under conditions that the objective functions are strong convex, and have Lipschitz continuous gradients, we manifest that proposed algorithm faster linearly converges to the global optimization solution than other algorithms as long as the chosen step sizes do not exceed an exact characterized upper bound. Numerical experiments are also provided to testify the theoretical analysis.

关键词： Optimization Convergence Acceleration Directed graphs Convex functions Communication networks Upper bound

Another Two-Timescale Duplex Neurodynamic Approach to Portfolio Selection 11

学校读者我要写书评

暂无评论

Another Two-Timescale Duplex Neurodynamic Approach to Portfo...

11th International Conference on intelligent Control and information processing, ICICIP 2021

作者： Leung, Man-Fai Wang, Jun Che, Hangjun Hong Kong Metropolitan University School of Science and Technology Hong Kong City University of Hong Kong Department of Computer Science School of Data Science Hong Kong Southwest University Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Chongqing China

ISBN: (纸本)9781665425155

This paper is concerned with portfolio selection based on the Markowitz mean-variance framework using neurodynamic optimization. The portfolio optimization problem is formulated as a biconvex optimization problem. A two-timescale duplex neurodynamic approach is then applied for solving the profolio selection problem. The approach makes use of two recurrent neural networks (RNNs) which operate at different timescales for local search. A particle swarm optimization algorithm is employed to update the neuronal states of the two RNNs for global optima. Experimental results on four stock market datasets show the superior performance of the neurodynamic approach in terms of long-term expected returns. © 2021 IEEE.

关键词： Recurrent neural networks

Software and Hardware Synergy for Accelerated Plant Disease Identification

学校读者我要写书评

暂无评论

SSRN 2024年

作者： Wen, Hongxing Chen, Ling College of Electronic Information and Engineering Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing Southwest University Chongqing400715 China

Plant diseases are one of the main causes of reduced crop yields. Therefore, it is necessary to adopt timely and effective identification methods and take corresponding measures. Some physical and biological detection methods have been proposed by related researchers, but these methods require specialised techniques and expensive detection costs, and the number of technicians is very limited, so the identification of plant diseases is always not timely and effective. In order to achieve real-time identification of plant diseases in some remote areas where networks and circuits are not well connected. We adopted a hardware and software co-acceleration approach to implement the design. First, we designed a lightweight convolutional neural network (CNN) and trained this network using a knowledge distillation approach. Then, we quantified the model parameters into int8 type using a method of model quantization to further compress the model size. After compression, the model size of the network is 0.035 MB and the recognition accuracy is 94.06% on the test set of the experiment. In order to deploy the proposed network on resource constrained FPGA devices, we used time-division multiplexing and feature map segmentation to deploy the network. Finally, our design is implemented on ZYNQ7020 with an inference speed of 35.73ms/frame and a power consumption of 1.97 W. The experimental results show that our design has the advantages of consuming less FPGA resources, low power consumption, high speed, and portability, and can be used for disease recognition in multiple plant classes. © 2024, The Authors. All rights reserved.

关键词： Field programmable gate arrays (FPGA)

Kernel Least Mean Square Based on the Sparse Nyström Method

学校读者我要写书评

暂无评论

Kernel Least Mean Square Based on the Sparse Nyström Method

IEEE International Symposium on circuits and Systems (ISCAS)

作者： Haonan Zhang Han Jiang Shiyuan Wang Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Southwest University Chongqing 400715 China

ISBN: (数字)9781728133201

ISBN: (纸本)9781728133218

Kernel least mean square based on the Nyström method (NysKLMS) has been proposed to fix the network structure of kernel least mean square (KLMS) by approximating a large Gram matrix with a low-rank matrix generated by sampling from input vectors. However, the computational burden of NysKLMS increases with the growth of the input dimension. To alleviate this computational burden, a novel KLMS based on a sparse Nyström method (SNKLMS) algorithm is proposed in this paper. Unlike NysKLMS using the k-means sampling to obtain k centroids from input data, the proposed SNKLMS algorithm first divides input data into several clusters, and then selects k centroids from the obtained clusters equally. According to the relation between the Gaussian kernel and the distance of different clusters, partial submatrices in the low-rank matrix of SNKLMS can be omitted to further reduce the operations of multiplication and addition, and thus a sparse low-rank matrix is obtained. Simulations on the channel equalization and chaotic time-series prediction validate the superiorities of SNKLMS.

关键词： Kernel Sparse matrices Signal processing algorithms Approximation algorithms Clustering algorithms Matrix decomposition Training data

Finite-Time Stability of Solutions for Non-Instantaneous Impulsive Systems and Application to Non-Instantaneous Impulsive Neural Networks

学校读者我要写书评

暂无评论

SSRN 2022年

作者： Deng, Hao Li, Chuandong Wang, Yinuo Wu, Hongjuan Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing College of Electronic and Information Engineering Southwest University Chongqing400715 China

This work extends the finite-time stability (FTS) results to non-instantaneous impulsive time-varying systems on the basis of general impulsive systems via Lyapunov theory. The impacts of impulse numbers, impulse time sequences and impulse amplitudes on FTS and settling-time estimation are discussed. The effects of different types of impulses on settling-time are discussed when the impulse number is known or pre-given. The average impulsive interval is introduced to extend the above results to the case when the impulse number is unknown, and the derivative of Lyapunov function is allowed to be indefinite instead of being negative definite or semi-negative definite. Moreover, the theoretical results are applied to non-instantaneous impulsive neural networks. Finally, we provide two numerical examples to demonstrate the effectiveness and feasibility of the obtained results. © 2022, The Authors. All rights reserved.

关键词： Lyapunov functions