检索结果-内蒙古大学图书馆

16th International Conference on Communication software and Networks, ICCSN 2024

作者： Wang, Hongliang Wang, Kaizhu Ji, Binglong Deng, Changyi Shenyang Institute of Computing Technology Chinese Academy of Sciences University of Chinese Academy of Sciences Liaoning Province Human- Computer Interaction System Engineering Research Center Based on DigitTwin Shenyang China Shenyang Institute of Computing Technology Chinese Academy of Sciences University of Chinese Academy of Sciences Shenyang China Institute of Software China Industrial Control Systems Cyber Emergency Response Team Beijing China

ISBN: (纸本)9798331506322

In the production and assembly process, the accuracy of detection is usually low due to limited resources and small quantities. In this paper, we propose a few-shot learning method for object detection based on the combination of YOLOv8 and Mamba-Like Linear Attention (MLLA) module. This method utilizes a high-pixel camera to capture the shape and size of the workpiece. By consistency-based mutual augmentation structure for few-shot samples data augmentation. By introducing the efficient and lightweight MLLA module, the model's speed and accuracy in object detection and segmentation tasks on mobile devices are significantly improved. Specifically, MLLA mainly integrates the two key elements of 'forget gate' and 'block design' in Mamba, and MLLA replaces forget gate by using position coding (RoPE) to provide necessary location information while maintaining parallel computation and fast inference speed. This makes MLLA improves the accuracy of detection and segmentation recognition. Experimental results show that compared to the original YOLOv8 model, the model with the MLLA module exhibits improvements in segmentation precision and recognition accuracy. This method demonstrates high accuracy and robustness in workpiece recognition, making it suitable for real-time industrial applications. © 2024 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

FGLight: Learning Neighbor-level Information for Traffic Signal Control 25

FGLight: Learning Neighbor-level Information for Traffic Sig...

引用

Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems

作者： Hang Xiao Huale Li Shuhan Qi Jiajia Zhang DingZhong Cai School of Software Northwest Polytechnical University Xi'an China School of Software Northwest Polytechnical University & Yangtze River Delta Research Institute of NPU Xi 'an China School of Computer Science and Technology Harbin Insitute of Technology Shenzhen Shenzhen China School of Computer Science and Technology Harbin Institute of Technology Shenzhen Shenzhen China School of Software Northwestern Polytechnical University Xi'an China

ISBN: (纸本)9798400714269

In recent years, multi-agent reinforcement learning (MARL) methods have increasingly been applied to traffic signal control and have achieved some success. However, most of existing MARL methods often underemphasize the heterogeneity in neighborhood-level information of the same agent. This results in highly sensitive performances and a long learning process. To address this challenge, we propose FGLight, a novel Feudal MARL method for traffic signal control. FGLight leverages Adaptive Graph Attention Networks (AGAT) to dynamically model the interactive relationships between intersections. Through adaptive neighbor selection and weight-based attention mechanisms, AGAT dynamically assigns importance weights to neighbor-level information, thereby improving the accuracy of local policies by more effectively exploiting neighborhood information. Moreover, FGLight introduces a Smooth Hysteretic Deep Q-Network (SHDQN) based on an optimistic assumption mechanism, which enhances the stability of the global policy. We conducted experiments on both synthetic and real-world datasets, and the results demonstrate that, compared to several state-of-the-art MARL methods, FGLight performs better as the complexity of the road network increases, exhibiting faster convergence and greater policy stability.

关键词： graph attention networks

来源：评论

学校读者我要写书评

暂无评论

PathLAD+: An Improved Exact Algorithm for Subgraph Isomorphism Problem 32

PathLAD+: An Improved Exact Algorithm for Subgraph Isomorphi...

引用

32nd International Joint Conference on Artificial Intelligence, IJCAI 2023

作者： Wang, Yiyuan Jin, Chenghou Cai, Shaowei Lin, Qingwei School of Computer Science and Information Technology Northeast Normal University China Key Laboratory of Applied Statistics of MOE Northeast Normal University Changchun China Computer School Beijing Information Science and Technology University Beijing China State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China School of Computer Science and Technology University of Chinese Academy of Sciences China Microsoft Research China

ISBN: (纸本)9781956792034

The subgraph isomorphism problem (SIP) is a challenging problem with wide practical applications. In the last decade, despite being a theoretical hard problem, researchers designed various algorithms for solving SIP. In this work, we propose three main heuristics and develop an improved exact algorithm for SIP. First, we design a probing search procedure to try whether the search procedure can successfully obtain a solution at first sight. Second, we design a novel matching ordering as a value-ordering heuristic, which uses some useful information obtained from the probing search procedure to preferentially select some promising target vertices. Third, we discuss the characteristics of different propagation methods in the context of SIP and present an adaptive propagation method to make a good balance between these methods. Experimental results on a broad range of real-world benchmarks show that our proposed algorithm performs better than state-of-the-art algorithms for the SIP. © 2023 International Joint Conferences on Artificial Intelligence. All rights reserved.

关键词： Set theory

来源：评论

学校读者我要写书评

暂无评论

Low-rank matrix recovery with total generalized variation for defending adversarial examples

引用

Frontiers of Information technology & Electronic Engineering 2024年第3期25卷 432-445页

作者： Wen LI Hengyou WANG Lianzhi HUO Qiang HE Linlin CHEN Zhiquan HE Wing W.Y.Ng School of Science Beijing University of Civil Engineering and ArchitectureBeijing 100044China School of Computer Science and Engineering South China University of TechnologyGuangzhou 510006China Aerospace Information Research Institute Chinese Academy of SciencesBeijing 100094China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen 518060China Institute of Big Data Modeling and Technology Beijing University of Civil Engineering and ArchitectureBeijing 100044China

Low-rank matrix decomposition with first-order total variation(TV)regularization exhibits excellent performance in exploration of image *** advantage of its excellent performance in image denoising,we apply it to improve the robustness of deep neural ***,although TV regularization can improve the robustness of the model,it reduces the accuracy of normal samples due to its *** our work,we develop a new low-rank matrix recovery model,called LRTGV,which incorporates total generalized variation(TGV)regularization into the reweighted low-rank matrix recovery *** the proposed model,TGV is used to better reconstruct texture information without *** reweighted nuclear norm and Li-norm can enhance the global structure ***,the proposed LRTGV can destroy the structure of adversarial noise while re-enhancing the global structure and local texture of the *** solve the challenging optimal model issue,we propose an algorithm based on the alternating direction method of *** results show that the proposed algorithm has a certain defense capability against black-box attacks,and outperforms state-of-the-art low-rank matrix recovery methods in image restoration.

关键词： Total generalized variation Low-rank matrix Alternating direction method of multipliers Adversarial example

来源：评论

学校读者我要写书评

暂无评论

Overview of Medical Data Privacy Protection based on Blockchain technology 7

Overview of Medical Data Privacy Protection based on Blockch...

引用

7th IEEE International Conference on Smart Cloud, SmartCloud 2022

作者： Chen, Lin Yu, Qingchun Liang, Wei Cai, Jiahong Zhu, Hangyu Xie, Songyou Hunan University of Science and Technology Hunan Key Laboratory for Service computing and Novel Software Technology School of Computer Science and Engineering Xiangtan China Shenzhen Research Institute Sun Yat-sen University and Novel Software Technology ShenZhen China Hunan University College of Computer Science and Electronic Engineering Changsha China

ISBN: (数字)9781665451796

ISBN: (纸本)9781665451796

With the increasing growth of electronic medical data, the difficulties of data sharing among medical institutions and the leakage of data privacy have become the focus of the public and medical workers. The blockchain has the characteristics of decentralization, traceability, and immutability, which can provide new ideas for fine-grained secure access to medical research. This article first introduces blockchain and blockchain-based privacy protection technology;then analyzes the advantages and disadvantages of electronic medical records, and introduces the current development status of electronic medical records based on blockchain technology;then from data encryption, access the three aspects of control and transaction anonymity introduce the medical data privacy protection method based on blockchain technology;finally, the full text is summarized and prospected. © 2022 IEEE.

关键词： Blockchain

来源：评论

学校读者我要写书评

暂无评论

Digital Human Techniques for Education Reform 24

Digital Human Techniques for Education Reform

引用

7th International Conference on Educational technology Management, ICETM 2024

作者： Guo, Peirong Zhang, Qi Tian, Chunwei Xue, Wanli Feng, Xiaocheng School of Software Northwestern Polytechnical University Shaanxi Xi’an China School of Economics and Management Harbin Institute of Technology at Weihai Shandong Weihai China Shenzhen Research Institute of Northwestern Polytechnical University Northwestern Polytechnical University Guangdong Shenzhen China School of Computer Science and Engineering Tianjin University of Technology Tianjin China Research Center for Social Computing and Information Retrieval Harbin Institute of Technology Heilongjiang Harbin China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security Shenzhen University Guangdong Shenzhen China

ISBN: (纸本)9798400717468

The rapid evolution of artificial intelligence, big data, and generative AI models has ushered in significant transformations across various sectors, including education. Digital Human Technique, an innovative technology grounded in advanced computer science and artificial intelligence, is reshaping educational paradigms by enabling virtual humans to simulate human behavior, express emotions, and interact with users. This paper explores the application of Digital Human Technique in education reform, focusing on creating immersive, intelligent classroom experiences that foster meaningful interactions between teachers and students. We define Digital Human Technique and delve into its key technical components such as character modeling and rendering, natural language processing, computer vision, and augmented reality technologies. Our methodology involves analyzing the role of educational digital humans created through these technologies, assessing their impact on educational processes, and examining various application scenarios in educational reform. Results indicate that Digital Human Technique significantly enhances the learning experience by enabling personalized teaching, increasing engagement, and fostering emotional connections. Educational digital humans serve as virtual teachers, interactive learning aids, and facilitators of emotional interaction, effectively addressing the challenges of traditional educational methods. They also promote a deeper understanding of complex concepts through simulated environments and interactive digital content. © 2024 Copyright held by the owner/author(s).

关键词： Digital elevation model

来源：评论

学校读者我要写书评

暂无评论

3rd Workshop on Adverse Impacts and Collateral Effects of Artificial Intelligence Technologies

引用

CEUR Workshop Proceedings 2023年 3593卷

作者： Aïmeur, Esma Díaz Ferreyra, Nicolás E. Department of Computer Science and Operations Research University of Montréal Canada Institute of Software Security Hamburg University of Technology Germany

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Machine Unlearning 1

引用

17th International Conference on Learning and Intelligent Optimization, LION-17 2023

作者： Zhu, HongBin Xia, YuXiao Li, YunZhao Li, Wei Liu, Kang Gao, Xianzhou Big Data Center State Grid of China Beijing100012 China State Grid Smart Grid Research Institute Shanghai200016 China State Key Laboratory for Novel Software Technology Department of Computer Science and Technology Nanjing University Nanjing200016 China

ISBN: (数字)9783031445057

ISBN: (纸本)9783031445040

In recent years, deep neural networks have enjoyed tremendous success in industry and academia, especially for their applications in visual recognition and natural language processing. While large-scale deep models bring incredible performance, their massive data requirements pose a huge threat to data privacy protection. With the growing emphasis on data security, the study of data privacy leakage in machine learning, such as machine unlearning, has become increasingly important. There have been many works on machine unlearning, and other research has proposed training several submodels to speed up the retraining process, by dividing the training data into several disjoint fragments. When the impact of a particular data point in the model is to be removed, the model owner simply retrains the sub-model containing this data point. Nevertheless, current learning methods for machine unlearning are still not widely used due to model applicability, usage overhead, etc. Based on this situation, we propose a novel hierarchical learning method, Hierarchical Machine Unlearning (HMU), with the known distribution of unlearning requests. Compared with previous methods, ours has better efficiency. Using the known distribution, the data can be partitioned and sorted, thus reducing the overhead in the data deletion process. We propose to train the model using the hierarchical data set after partitioning, which further reduces the loss of prediction accuracy of the existing methods. It is also combined with incremental learning methods to speed up the training process. Finally, the effectiveness and efficiency of the method proposed in this paper are verified by multiple experiments. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Data privacy

来源：评论

学校读者我要写书评

暂无评论

Towards Query-Efficient Decision-Based Adversarial Attacks Through Frequency Domain

Towards Query-Efficient Decision-Based Adversarial Attacks T...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Jianhao Fu Xiang Ling Yaguan Qian Changjiang Li Tianyue Luo Jingzheng Wu Intelligent Software Research Center Institute of Software Chinese Academy of Sciences University of Chinese Academy of Sciences Nanjing Nanjing Institute of Software Technology Key Laboratory of System Software (Chinese Academy of Sciences) State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Zhejiang University of Science and Technology Stony Brook University

ISBN: (数字)9798350390155

ISBN: (纸本)9798350390162

Deep neural networks are vulnerable to adversarial examples, where decision-based attacks can generate adversarial examples based solely on the predicted labels. However, these attacks typically require excessive queries to attack one example. Considering this challenge, we propose FBA (Frequency based Boundary Attack), a decision-based attack against the limitation of query efficiency. FBA incorporates a novel search process, utilizing high-frequency based importance sampling for efficient gradient estimation. Empirical results confirm the superior query efficiency of our method. Specifically, FBA surpasses SOTA attacks by achieving a 54% average improvement in query efficiency, quantified by the reduction in perturbation size within the same number of queries.

关键词： Technological innovation Monte Carlo methods Frequency-domain analysis Perturbation methods Estimation Artificial neural networks

来源：评论

学校读者我要写书评

暂无评论

Patch Target Guided Dual-Branch Deep Multiple Instance Learning for 3D MRI Analysis

Patch Target Guided Dual-Branch Deep Multiple Instance Learn...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Dai, Mingzhe Shi, Xiaoshuang Zhu, Xiaofeng Pan, Tingrui Li, Kang University of Science and Technology of China School of Software Engineering Hefei China West China Hospital of Sichuan University West China Biomedical Big Data Center Chengdu China University of Electronic Science and Technology of China School of Computer Science and Engineering Chengdu China Sichuan Artificial Intelligence Research Institute Yibin China University of Science and Technology of China Suzhou Institute for Advanced Research Suzhou China Shanghai Artificial Intelligence Laboratory Shanghai China

ISBN: (纸本)9798350386226

Deep multiple instance learning (MIL) has attracted considerable attention in medical image analysis, since it only requires image-level labels for model training without using fine-grained (or patch) annotations. Unfortunately, MIL-based methods might lose some significant patch features. Although pseudo-label-based methods, which assign a pre-defined label to each patch, can explore more patch-level features, they might bring label noise and make the patch-level features lose diversity, thereby possibly restricting the model performance. To overcome this issue, we propose a novel gradient-based patch target generation (PTG) module to dynamically produce a feature vector for each patch as its target. Additionally, based on the PTG module, we propose a patch-target guided dual-branch deep MIL framework for 3D MRI data analysis, where both the two branches consist of a CNN model to extract patch-level features, an attention module to interpret the significance of patches, and a bag-level classifier, while the second branch also contains the PTG module to generate patch targets of patches. Moreover, the two branches are alternatively updated in our framework, resulting in a bi-level optimization problem, and thus we design a bi-level optimization algorithm to solve our proposed objective function. Extensive experiments demonstrate the superior classification and interpretation performance of the proposed framework over recent state-of-the-art methods. Codes are available at https://***/daimz1213/mil2024. © 2024 IEEE.

关键词： Image annotation

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：