检索结果-内蒙古大学图书馆

8th International Conference on Learning Representations, ICLR 2020

作者： Zhang, Xiao Wu, Dongrui Ministry of Education Key Laboratory of Image Processing and Intelligent Control School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan China

A deep neural network (DNN) with piecewise linear activations can partition the input space into numerous small linear regions, where different linear functions are fitted. It is believed that the number of these regions represents the expressivity of the DNN. This paper provides a novel and meticulous perspective to look into DNNs: Instead of just counting the number of the linear regions, we study their local properties, such as the inspheres, the directions of the corresponding hyperplanes, the decision boundaries, and the relevance of the surrounding regions. We empirically observed that different optimization techniques lead to completely different linear regions, even though they result in similar classification accuracies. We hope our study can inspire the design of novel optimization techniques, and help discover and analyze the behaviors of DNNs. © 2020 8th International Conference on Learning Representations, ICLR 2020. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Virtually-Federated Scheduling of Parallel Real-Time Tasks 42

Virtually-Federated Scheduling of Parallel Real-Time Tasks

引用

42nd IEEE Real-Time Systems Symposium, RTSS 2021

作者： Jiang, Xu Guan, Nan Liang, Haochun Tang, Yue Qiao, Lei Yi, Wang Key Laboratory of Intelligent Computing in Medical Image Ministry of Education Northeastern University China City University of Hong Kong Hong Kong Beijing Institute of Control Engineering China Uppsala University Sweden

ISBN: (纸本)9781665428026

Federated scheduling is a promising approach to schedule parallel real-time tasks, where each task exclusively executes on a set of dedicated processors. However, federated scheduling suffers significant resource wasting since a task typically only uses part of the processing capacity allocated to it, while the unused part cannot be shared with other tasks. To solve this problem, we present a virtually-federated scheduling approach, which both enjoys the good analyzability of federated scheduling and allows tasks to efficiently share processors with others. The main idea is to construct virtual processors on physical processors, and let a task exclusively execute on a set of virtual processors. As a physical processor is shared by virtual processors, tasks effectively share processors with each other. On the other hand, as each task exclusively executes on its own virtual processor set, the good analyzability of federated scheduling can be carried into to our virtually-federated scheduling approach. We conduct comprehensive performance evaluation to compare our proposed approach with existing methods of different types. Experiment results show that our approach consistently outperforms existing methods to a considerable extent under a wide range of parameter settings. ©2021 IEEE

关键词： Scheduling

来源：评论

学校读者我要写书评

暂无评论

Generative Adversarial Network-Based Electromagnetic Signal Classification: A Semi- Supervised Learning Framework

引用

China Communications 2020年第10期17卷 157-169页

作者： Huaji Zhou Licheng Jiao Shilian Zheng Lifeng Yang Weiguo Shen Xiaoniu Yang Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education School of Artificial IntelligenceXidian UniversityXi'an 710071China Science and Technology on Communication Information Security Control Laboratory Jiaxing 314033China

Generative adversarial network(GAN)has achieved great success in many fields such as computer vision,speech processing,and natural language processing,because of its powerful capabilities for generating realistic *** this paper,we introduce GAN into the field of electromagnetic signal classification(ESC).ESC plays an important role in both military and civilian ***,in many specific scenarios,we can’t obtain enough labeled data,which cause failure of deep learning methods because they are easy to fall into ***,semi-supervised learning(SSL)can leverage the large amount of unlabeled data to enhance the classification performance of classifiers,especially in scenarios with limited amount of labeled *** present an SSL framework by incorporating GAN,which can directly process the raw in-phase and quadrature(IQ)signal *** to the characteristics of the electromagnetic signal,we propose a weighted loss function,leading to an effective classifier to realize the end-to-end classification of the electromagnetic *** validate the proposed method on both public RML2016.04c dataset and real-world Aircraft Communications Addressing and Reporting System(ACARS)signal *** experimental results show that the proposed framework obtains a significant increase in classification accuracy compared with the state-of-the-art studies.

关键词： generative adversarial network semi-supervised learning electromagnetic signal classification end-to-end classification weighted loss function

来源：评论

学校读者我要写书评

暂无评论

Parameter training methods for convolutional neural networks with adaptive adjustment method based on Caputo fractional-order differences

引用

Chaos, Solitons and Fractals 2025年 199卷

作者： Zhao, Haiming Yang, Honggang Chen, Jiejie Jiang, Ping Zeng, Zhigang School of Computer and Information Engineering Hubei Normal University Huangshi435000 China School of Automation Wuhan University of Technology Wuhan430000 China School of Computer Hubei Polytechnic University Huangshi435000 China School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430000 China Key Laboratory of Image Information Processing and Intelligent Control Ministry of Education of China Wuhan430000 China

As deep learning technologies continue to permeate various sectors, optimization algorithms have become increasingly crucial in neural network training. This paper introduces two adaptive momentum algorithms based on Grünwald–Letnikov and Caputo fractional-order differences—Fractional Order Adagrad (FAdagrad) and Fractional Order Adam (FAdam)—to update parameters more flexibly by adjusting momentum information. Commencing from the definitions of fractional derivatives, we propose integrating fractional-order differences with gradient algorithms in convolutional neural networks (CNNs). These adaptive momentum algorithms, leveraging Grünwald–Letnikov and Caputo fractional-order differences, offer enhanced flexibility, thereby accelerating convergence. Our nonlinear parameter tuning method for CNNs demonstrates superior performance compared to traditional integer-order momentum algorithms and the standard Adam algorithm. Experimental results on the BraTS2021 dataset and CIFAR-100 dataset reveal that the proposed fractional-order optimization algorithms significantly outperform their integer-order counterparts in model optimization. They not only expedite convergence but also improve the accuracy of image recognition and segmentation. © 2025 Elsevier Ltd

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

A Single-Cell Clustering Algorithm Based on Structure Perturbation Non-Negative Matrix Factorization

A Single-Cell Clustering Algorithm Based on Structure Pertur...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Jiang, Hanjing Wang, Meineng Zhang, Luping Huang, Yu-An Li, Yangyuan Huang, Yabing Huazhong University of Science and Technology Key Laboratory of Image Information Processing and Intelligent Control Education Ministry of China Institute of Artificial Intelligence School of Artificial Intelligence and Automation Hubei Wuhan430074 China China University of Mining and Technology School of Computer Science and Technology Xuzhou221116 China Yichun University School of Mathematics and Computer Science Yichun336000 China East China University of Technology School of Information Engineering Nanchang330013 China Northwestern Polytechnical University School of Computer Science Xi'an710129 China Renmin Hospital Wuhan University Department of Pathology Wuhan430060 China

ISBN: (纸本)9798350386226

The advent of single-cell RNA sequencing (scRNA-seq) has facilitated the acquisition of high-resolution data regarding cell heterogeneity across various tissues. A fundamental and critical step in the analysis of scRNA-seq data is cell type identification. An appropriate graph construction strategy can reflect the topological structure between cells and allows for detailed exploration of the relationships between cells and genes. Given the sparsity of scRNA-seq data and the pronounced noise generated by shallow sequencing, choosing an appropriate graph construction strategy has become a significant challenge. In this paper, we propose a single-cell clustering method, named Sc-PNNMF, based on structural perturbation of nonnegative matrix factorization. Sc-PNNMF employs a structural perturbation algorithm to optimize the gene expression matrix. The optimized matrix guides the matrix factorization process, effectively overcoming the impact of data sparsity and significant noise in gene expression data on the graph construction strategy. We compared the clustering abilities of Sc-PNNMF with five other methods on ten real datasets. © 2024 IEEE.

关键词： Non-negative matrix factorization

来源：评论

学校读者我要写书评

暂无评论

Object relationship graph reasoning for object detection of remote sensing images 6

Object relationship graph reasoning for object detection of ...

引用

6th International Conference on image, Vision and Computing, ICIVC 2021

作者： Li, Zeng Liu, Yifan Liu, Jingdong Yuan, Ye Raza, Asif Huo, Hong Fang, Tao Department of Automation Shanghai Jiao Tong University Shanghai China Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai China

ISBN: (纸本)9781665443685

In recent years, deep learning has been successfully applied in object detection of remote sensing images, due to its powerful feature extraction and representation capabilities. However, it usually ignores the relationships among different objects, which greatly hinders the further improvement of the detection precision. In this paper, a novel object relationship graph reasoning deep learning method is proposed for object detection of remote sensing images. The reasoning is carried out on a graph constructed by region proposals. It mainly contains three modules for relationship generation in consideration of different factors that related to object detection: 1) Location-based relationship generation module (LRGM) aims to find spatial relationships among objects according to their positions;2) Feature-based relationship generation module (FRGM) focuses on mining semantic relationships among objects according to their appearances that implied in feature maps and 3) Class-dependent relationship generation module (CRGM) introduces class dependencies into the relationship generation process. Class dependencies are the prior knowledge that is obtained by computing class co-occurrence. The experimental results on a public twenty-class remote sensing object detection dataset have shown that the proposed method has better detection performance compared to other object detection deep networks with no considering object relationships. © 2021 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Learning from Noisy Labels with Coarse-to-Fine Sample Credibility Modeling

arXiv

引用

arXiv 2022年

作者： Zhang, Boshen Li, Yuxi Tu, Yuanpeng Peng, Jinlong Wang, Yabiao Wu, Cunlin Xiao, Yang Zhao, Cairong YouTu Lab Tencent Shanghai China Tongji University Shanghai China Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology China

Training deep neural network (DNN) with noisy labels is practically challenging since inaccurate labels severely degrade the generalization ability of DNN. Previous efforts tend to handle part or full data in a unified denoising flow via identifying noisy data with a coarse small-loss criterion to mitigate the interference from noisy labels, ignoring the fact that the difficulties of noisy samples are different, thus a rigid and unified data selection pipeline cannot tackle this problem well . In this paper, we first propose a coarse-to-fine robust learning method called CREMA, to handle noisy data in a divide-and-conquer manner. In coarse-level, clean and noisy sets are firstly separated in terms of credibility in a statistical sense. Since it is practically impossible to categorize all noisy samples correctly, we further process them in a fine-grained manner via modeling the credibility of each sample. Specifically, for the clean set, we deliberately design a memory-based modulation scheme to dynamically adjust the contribution of each sample in terms of its historical credibility sequence during training, thus alleviating the effect from noisy samples incorrectly grouped into the clean set. Meanwhile, for samples categorized into the noisy set, a selective label update strategy is proposed to correct noisy labels while mitigating the problem of correction error. Extensive experiments are conducted on benchmarks of different modality, including image classification (CIFAR, Clothing1M etc) and text recognition (IMDB), with either synthetic or natural semantic noises, demonstrating the superiority and generality of CREMA. © 2022, CC BY.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Multi-target tracking control system of phased array antenna with mechanical scanning and beam scanning method

Multi-target tracking control system of phased array antenna...

引用

Fully Actuated System Theory and Applications (CFASTA), Conference on

作者： Jijun Ma Dewei Li Lianghai Li Shibei Xue Yuning Wang Jun Jia Department of Automation Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai China Beijing Research Institute of Telemetry Beijing China

In the field of telemetry, track, and command system, this study presents a phased array antenna control system for achieving multi-target tracking tasks through a set of systems. Firstly, electric beam scanning of a planar array phased array antenna is used in this paper to implement multi-target tracking. In addition, the phased array antenna is mounted on the antenna pedestal, which can operate on the azimuth and elevation mechanical axes. Thirdly, the mechanical scanning method of the azimuth-elevation antenna pedestal and the beam scanning method of the phased array antenna are combined to realize multi-target tracking tasks. Finally, the experiments of flight tracking are completed by employing the technique of UAV tagging, and the results validate the viability of the system.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Differentiator-Based Adaptive H∞ Tracking control of Fully Actuated Systems

Differentiator-Based Adaptive H∞ Tracking Control of Fully ...

引用

Fully Actuated System Theory and Applications (CFASTA), Conference on

作者： Yuxin Feng Zhiqiang Li Yang Liu Zhaoshui He Hongyi Li School of Automation and Guangdong Provincial Key Laboratory of Intelligent Decision and Cooperative Control Guangdong University of Technology Guangzhou China School of Automation and Electronics Engineering Qingdao University of Science and Technology Qingdao China School of Automation Guangdong University of Technology Guangzhou China Key Laboratory for IoT Intelligent Information Processing and System Integration of Ministry of Education Guangzhou China

ISBN: (数字)9798350373691

ISBN: (纸本)9798350373707

This article addresses an adaptive tracking control problem for uncertain high-order fully actuated (HOFA) systems with unknown parameters and disturbances. Under the framework of backstepping, the unknown parameter is estimated and the external disturbance is handled by using adaptive control and H ∞ technique, respectively. Moreover, the Levant differentiator is employed to reduce the computation burden. Additionally, the boundedness of signals in the closed-loop system is proven using the Lyapunov theory. Finally, the effectiveness of the proposed scheme is validated through the simulation study.

关键词： Backstepping Manipulators Stability analysis Closed loop systems Adaptive control Lyapunov methods

来源：评论

学校读者我要写书评

暂无评论

CrossGLG: LLM Guides One-shot Skeleton-based 3D Action Recognition in a Cross-level Manner

arXiv

引用

arXiv 2024年

作者： Yan, Tingbing Zeng, Wenzheng Xiao, Yang Tong, Xingyu Tan, Bo Fang, Zhiwen Cao, Zhiguo Zhou, Joey Tianyi Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China School of Biomedical Engineering Southern Medical University Guangzhou510515 China Department of Rehabilitation Medicine Zhujiang Hospital Southern Medical University Guangzhou510280 China Singapore Singapore Singapore Singapore

Most existing one-shot skeleton-based action recognition focuses on raw low-level information (e.g., joint location), and may suffer from local information loss and low generalization ability. To alleviate these, we propose to leverage text description generated from large language models (LLM) that contain high-level human knowledge, to guide feature learning, in a global-local-global way. Particularly, during training, we design 2 prompts to gain global and local text descriptions of each action from an LLM. We first utilize the global text description to guide the skeleton encoder focus on informative joints (i.e.,global-tolocal). Then we build non-local interaction between local text and joint features, to form the final global representation (i.e., local-to-global). To mitigate the asymmetry issue between the training and inference phases, we further design a dual-branch architecture that allows the model to perform novel class inference without any text input, also making the additional inference cost neglectable compared with the base skeleton encoder. Extensive experiments on three different benchmarks show that CrossGLG consistently outperforms the existing SOTA methods with large margins, and the inference cost (model size) is only 2.8% than the previous SOTA. CrossGLG can also serve as a plug-and-play module that can substantially enhance the performance of different SOTA skeleton encoders with a neglectable cost during inference. The source code will be released soon. © 2024, CC BY.

关键词： Signal encoding

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：