检索结果-内蒙古大学图书馆

Data Driven Control and Learning Systems (DDCLS)

作者： Jian Zhang Xiaohe Qu Zhiwei Lei Haifeng Zhang Kai Zhong Datang East China Electric Power Test and Research Institute Hefei China Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Mathematical Science Anhui University Hefei China Key Laboratory of Intelligent Computing and Signal Processing of the Ministry of Education Institutes of Physical Science and Information Technology Anhui University Hefei Anhui China

The traditional kernel principal components analysis (KPCA) and linear discriminant analysis (LDA) have been verified to be two effective approaches for fault detection and diagnosis in recent years. Nevertheless, the conventional method and corresponding improved ones still exposed their deficiencies in some ways. Facing this dilemma, this paper presents a combination of optimized KPCA and modified LDA (OKPCA-MLDA), in which the OKPCA avoids the loss of original features after centralizing data in the eigenspace by adjusting covariance matrix's eigenvalue and transforming the distribution of variables thus providing representative and abundant principal components for the classifier. In addition, the MLDA maximizes a brand new objective function in feature space which achieves better classification performance than the conventional LDA, then utilizing diagnostic thresholds and similarity coefficients to identify the fault types. Based on the combined model, not only the fault detection and diagnosis can be realized simultaneously but also the accuracy of detection and diagnosis can be guaranteed. Furthermore, the simulation experiments on Tennessee Eastman (TE) benchmark process clearly illustrated the superiority of our proposed strategy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Multi-Objective Neural Architecture Search for Generalized Cognitive Diagnosis Models

Evolutionary Multi-Objective Neural Architecture Search for ...

引用

Data-driven Optimization of Complex Systems (DOCS), International Conference on

作者： Shangshang Yang Cheng Zhen Ye Tian Haiping Ma Yuanchao Liu Panpan Zhang Xingyi Zhang Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Artificial Intelligence Anhui University Hefei China Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Institutes of Physical Science and Information Technology Anhui University Hefei China State Key Laboratory of Synthetical Automation for Process Industries Northeastern University Shenyang China

Cognitive diagnosis models (CDMs) with high generalization are essential for intelligent education systems to reveal students' knowledge states in multiple datasets. However, existing CDMs' architectures are designed dependent on researcher expertise and experience from observing and summarizing partial students' learning behaviors, which makes handcrafted models not cover all learning behaviors well and thus limits their generalization. To develop generalized CDMs, this paper proposes an evolutionary neural architecture search to design CDMs' architectures effective on multiple datasets automatically. Specifically, we first formulate the search task as a multi-objective optimization problem (MOP), which maximizes model performance on multiple datasets containing learning behaviors as many as possible to ensure model generalization. Then, an expressive search space is devised to contain as many potential architectures as possible, where each architecture is denoted by a unified form, taking three given inputs and integrating them in a linear or no-linear manner for prediction. Finally, an evolutionary algorithm with a tailored deduplication strategy is employed to solve the MOP, where each architecture is further represented by a single-root computation tree for easy optimization. Experiments on multiple datasets demonstrate the generalization and effectiveness of the best architecture searched by the proposed approach, and the searched architecture also holds as good interpretability as handcrafted architectures.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SGCN: A Scalable Graph Convolutional Network with Graph-Shaped Kernels and Multi-Channels

SSRN

引用

SSRN 2023年

作者： Huang, Zhenhua Zhou, Wenhao Li, Kunhao Jia, Zhaohong School of Internet Anhui University Feixi Road Hefei230039 China Key Lab of Intelligent Computing and Signal Processing of Ministry of Education Hefei China

Graph neural networks (GNNs) have shown great success in graph processing. However, current message-passing-based GNNs have limitations in feature aggregations and update mechanisms that rely on a fixed mode, resulting in inadequate representations of the neighborhood structure's richness. Furthermore, the convolution layer in most GNNs lacks flexibility and multi-channels compared to convolutional neural networks (CNNs). CNNs employ multiple channels, where different kernels capture shared patterns of images. To address the limitations, we propose a novel scalable graph convolution network (SGCN). SGCN enhances structural information learning by introducing an expressive feature aggregation mechanism. Drawing inspiration from the design of CNNs, SGCN designs graph-shaped kernels to perform multi-channel convolutions on the subgraph structure. A sampler based on degree centrality is employed to simplify the computation costs. Extensive experiments demonstrate that SGCN achieved state-of-the-art performances on various scales of graph datasets (accuracy improved up to 5.38%). © 2023, The Authors. All rights reserved.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

A Spectrum Based Algorithm for Image Classification

引用

Chinese Journal of Electronics 2023年第3期18卷 427-430页

作者： Nian Wang Jiang Zhang Jun Tang Yizheng Fan Dong Liang Key Laboratory of Intelligent Computing and Signal Processing Anhui University Education Ministry Hefei China Hefei Teachers College Hefei China Department of Mathematics Anhui University Hefei China

In this paper, a novel algorithm for image classification is presented which uses the projective value of adjacency spectrum as classified samples. Firstly, the eigenvalues of adjacency matrices constructed on the feature point-sets of images are obtained by singular value decomposition. Secondly, the eigenvalues are projected onto the eigenspace by means of the covariance matrix. Finally, image classification is performed by adopting RBF and PNN neural networks as classifiers respectively. Mean-while, some theoretical analyses are given to support the proposed method.

关键词： Neural networks Education Eigenvalues and eigenfunctions Classification algorithms Matrix decomposition Covariance matrices Image classification

来源：评论

学校读者我要写书评

暂无评论

Sparse Andrew's Sine Norm Promoting Adaptive Algorithm under Non-Gaussian Noises 5

Sparse Andrew's Sine Norm Promoting Adaptive Algorithm under...

引用

5th IEEE International Conference on Electronic Information and Communication Technology, ICEICT 2022

作者： Hadi, Abdul Huang, Xinqi Ali, Burhan Li, Yingsong College of Information and Communications Engineering Harbin Engineering University Harbin150001 China College of Intelligent System and Control Science Engineering Harbin Engineering University Harbin150001 China Anhui University Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui Hefei China

ISBN: (数字)9781665472111

ISBN: (纸本)9781665472111

Two sparse adaptive filtering (AF) algorithms based on Andrew's sine estimator (ASE) are presented to achieve improved performance for identifying sparse systems, where the ASE is derived within the least-square framework. Furthermore, zero-attracting (ZA) scheme is used in ASE to construct ZA-ASE and its re-weighting form (RZA-ASE) to combat non-Gaussian noises and use the sparse characteristics of the system. Their performance is investigated via simulations and compared with the least-mean square (LMS) and the maximum correntropy criterion (MCC) algorithms to show their superior performance. © 2022 IEEE.

关键词： Adaptive filtering

来源：评论

学校读者我要写书评

暂无评论

Dual-Control Reconfigurable intelligent Metasurface: Design and Realization

Dual-Control Reconfigurable Intelligent Metasurface: Design ...

引用

IEEE International Conference on Electronic Information and Communication Technology (ICEICT)

作者： Kangtai Zheng Kwok L. Chung Yuzhong Li Guoming Lai Jianhai Wang Yingsong Li School of Computer Science and Engineering Huizhou University Huizhou China Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University China

A dual-control reconfigurable intelligent metasurface (DC-RIMS) composed of $24\times 22$ symmetric cells in the 5G mid-band is proposed. The DC-RIMS of non-artificial-magnetic-conductor type provides a smaller phase resolution of $60^{\mathrm{o}}$ with 2 bits under normal incidence. Angular sensitivities of the RIMS under oblique incidence are investigated. Our results show that the reflection responses at elevation angles are more sensitive than the azimuth angles, whereas the maximum phase differences are differed from each polarization. The horizontal one increases to $220^{\mathrm{o}}$ whereas the vertical one reduces to $150^\mathrm{o}$ as compared with overlapped $180^{\mathrm{o}}$ under normal incidence.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Machine learning prediction of network dynamics with privacy protection

引用

Physical Review Research 2022年第4期4卷 043076-043076页

作者： Xin Xia Yansen Su Linyuan Lü Xingyi Zhang Ying-Cheng Lai Hai-Feng Zhang The Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Mathematical Science Anhui University Hefei 230601 China

Predicting network dynamics based on data, a problem with broad applications, has been studied extensively in the past, but most existing approaches assume that the complete set of historical data from the whole network is available. This requirement presents a great challenge in applications, especially for large, distributed networks in the real world, where data collection is accomplished by many clients in a parallel fashion. Often, each client only has the time series data from a partial set of nodes, and the client has access to only partial time stamps of the whole set of time series data and the partial structure of the network. Due to privacy concerns or license-related issues, the data collected by different clients cannot be shared. Accurately predicting the network dynamics while protecting the privacy of different parties is a critical problem in modern times. Here, we propose a solution based on federated graph neural networks (FGNNs) that enables the training of a global dynamic model for all parties without data sharing. We validate the working of our FGNN framework through two types of simulations to predict a variety of network dynamics (four discrete and three continuous dynamics). As a significant real-world application, we demonstrate successful prediction of state-wise influenza spreading in the USA. Our FGNN scheme represents a general framework to predict diverse network dynamics through collaborative fusing of the data from different parties without disclosing their privacy.

关键词： Collective behavior in networks

来源：评论

学校读者我要写书评

暂无评论

Pctlfusion: A Progressive Fusion Network Via Contextual Texture Learning for Infrared and Visible Image

SSRN

引用

SSRN 2024年

作者： Lu, Yixiang Gong, Daiwei Zhao, Dawei Maksim, Davydau Gao, Qingwei The Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education School of Electrical Engineering and Automation Anhui University Hefei230601 China Belarusian State University of Informatics and Radioelectronics Minsk Belarus

The objective of infrared and visible image fusion is to generate a fused image that contains rich texture details and salient targets. However, most of the existing fusion methods tend to focus on preserving texture details in visible image and salient targets in infrared image, ignoring the fact that infrared image have richer texture details and better visual effects than visible image under low-light conditions. In other words, differences in information sources under the condition of illumination imbalance is ignored in setting the task goal. In addition, existing network designs usually ignore multimodal information interactions during the coding phase, which may result in loss of information. In this work, we propose a progressive fusion network via contextual texture learning for infrared and visible image, termed as PCTLFusion. To enhance the limited texture information, we design a content and contextual texture learning module (CCTLM) to model the texture and content of both the source images, respectively. Then, an affine transform-based multimodal pre-fusion module (ATMPM) is designed to learn and pre-fuse the complementary and similar enhanced information, which achieves information interaction during encoding. To learn fully feature information using multilayers, we iterate CCTLM and ATMPM three times in the process of model forming. Finally, the fused image is reconstructed by decoding the output of the last ATMPM. Numerous experiments show that our proposed fusion algorithm is superior to the state-of-the-art existing methods in both subjective and objective evaluations. © 2024, The Authors. All rights reserved.

关键词： Textures

来源：评论

学校读者我要写书评

暂无评论

A Novel Broadband EBG Structure For Simultaneous Switching Noise Suppression

A Novel Broadband EBG Structure For Simultaneous Switching N...

引用

2023 International Applied Computational Electromagnetics Society Symposium, ACES-Monterey 2023

作者： Sun, Guoxing Yan, Yeqiang Ren, Xingang Cai, Xueyuan Wu, Bo Ju, Huping Tao, Wei Wang, Gang Zou, Bao Huang, Zhixiang Wu, Xianliang Anhui Province Key Laboratory of Target Recognition and Feature Extraction Luan23700 China Engineering Technology Research Center of Specific Power Devices with Novel Semiconductor Materials Anhui Province Chizhou247100 China Key Laboratory of Intelligent Computing & Signal Processing Ministry of Education Anhui University Hefei230601 China

ISBN: (纸本)9781733509633

In this paper, a compact electromagnetic bandgap (EBG) structure for suppressing simultaneous switching noise (SSN) in high-speed circuits is proposed. At the bandwidth level of 0.35GHz to 20GHz, an ultra-wideband SSN suppression feature with a suppression depth of -40dB is achieved. The basic unit structure of the presented EBG is formed by etching an annular groove in the square patch and connecting the separated patches inside and outside the groove with a bridge adjacent to the end. An improved L-type bridge is used between the EBG units to further improve the EBG performance. Compared with the traditional L-type bridge, the suppression depth of the proposed electromagnetic bandgap structure is improved from -30dB to - 40dB. In addition, the time domain eye simulation results show that the structure has good signal integrity. © 2023 ACES.

关键词： Energy gap

来源：评论

学校读者我要写书评

暂无评论

Attacking Voice Anonymization Systems with Augmented Feature and Speaker Identity Difference

arXiv

引用

arXiv 2024年

作者： Zhang, Yanzhe Bi, Zhonghao Xiao, Feiyang Yang, Xuefeng Zhu, Qiaoxi Guan, Jian Group of Intelligent Signal Processing College of Computer Science and Technology Harbin Engineering University China Faculty of Computing Harbin Institute of Technology China Acoustics Lab University of Technology Sydney Australia

This study focuses on the First VoicePrivacy Attacker Challenge within the ICASSP 2025 signal processing Grand Challenge, which aims to develop speaker verification systems capable of determining whether two anonymized speech signals are from the same speaker. However, differences between feature distributions of original and anonymized speech complicate this task. To address this challenge, we propose an attacker system that combines Data Augmentation enhanced feature representation and Speaker Identity Difference enhanced classifier to improve verification performance, termed DA-SID. Specifically, data augmentation strategies (i.e., data fusion and SpecAugment) are utilized to mitigate feature distribution gaps, while probabilistic linear discriminant analysis (PLDA) is employed to further enhance speaker identity difference. Our system significantly outperforms the baseline, demonstrating exceptional effectiveness and robustness against various voice anonymization systems, ultimately securing a top-5 ranking in the challenge. Copyright © 2024, The Authors. All rights reserved.

关键词： Anonymity

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：