检索结果-内蒙古大学图书馆

12th International Conference on Learning Representations, ICLR 2024

作者： Ya, Mengxi Li, Yiming Dai, Tao Wang, Bin Jiang, Yong Xia, Shu-Tao Tsinghua Shenzhen International Graduate School Tsinghua University Shenzhen China The State Key Laboratory of Blockchain and Data Security Zhejiang University China ZJU-Hangzhou Global Scientific and Technological Innovation Center China College of Computer Science and Software Engineering Shenzhen University China Guangzhou Intelligence Communications Technology Co. Ltd. China

Saliency-based representation visualization (SRV) (e.g., Grad-CAM) is one of the most classical and widely adopted explainable artificial intelligence (XAI) methods for its simplicity and efficiency. It can be used to interpret deep neural networks by locating saliency areas contributing the most to their predictions. However, it is difficult to automatically measure and evaluate the performance of SRV methods due to the lack of ground-truth salience areas of samples. In this paper, we revisit the backdoor-based SRV evaluation, which is currently the only feasible method to alleviate the previous problem. We first reveal its implementation limitations and unreliable nature due to the trigger generalization of existing backdoor watermarks. Given these findings, we propose a generalization-limited backdoor watermark (GLBW), based on which we design a more faithful XAI evaluation. Specifically, we formulate the training of watermarked DNNs as a min-max problem, where we find the 'worst' potential trigger (with the highest attack effectiveness and differences from the ground-truth trigger) via inner maximization and minimize its effects and the loss over benign and poisoned samples via outer minimization in each iteration. In particular, we design an adaptive optimization method to find desired potential triggers in each inner maximization. Extensive experiments on benchmark datasets are conducted, verifying the effectiveness of our generalization-limited watermark. Our codes are available at https://***/yamengxi/GLBW. © 2024 12th International Conference on Learning Representations, ICLR 2024. 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

Deep Reinforcement Learning for Joint Service Placement and Request Scheduling in Mobile Edge Computing Networks

Deep Reinforcement Learning for Joint Service Placement and ...

引用

IEEE Symposium on computers and Communications (ISCC)

作者： Yuxuan Deng Xiuhua Li Chuan Sun Jinlong Hao Xiaofei Wang Victor C. M. Leung School of Big Data & Software Engineering Chongqing University Chongqing China Chongqing Changan Automobile Software Technology Co. Ltd Chongqing China TKLAN College of Intelligence & Computing Tianjin University Tianjin China College of Computer Science & Software Engineering Shenzhen University Shenzhen China Department of Electrical & Computer Engineering The University of British Columbia Vancouver Canada

Mobile edge computing aims to provide cloud-like services on edge servers located near Mobile Devices (MDs) with higher Quality of Service (QoS). However, the mobility of MDs makes it difficult to find a global optimal solution for the coupled service placement and request scheduling problem. To address these issues, we consider a three-tier MEC network with vertical and horizontal cooperation. Then we formulate the joint service placement and request scheduling problem in a mobile scenario with heterogeneous services and resource limits, and convert it into two Markov decision processes to decouple decisions across successive time slots. We propose a Cyclic Deep Q-network-based Service placement and Request scheduling (CDSR) framework to find a long-term optimal solution despite future information unavailability. Specifically, to solve the issue of enormous action space, we decompose the system agent and train them cyclically. Evaluation results demonstrates the effectiveness of our proposed CDSR on user-perceived QoS.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Correcting Noisy Multilabel Predictions: Modeling Label Noise through Latent Space Shifts

arXiv

引用

arXiv 2025年

作者： Huang, Weipeng Li, Qin Xiao, Yang Qiao, Cheng Cai, Tie Liang, Junwei Hurley, Neil J. Piao, Guangyuan School of Computer Science and Software Engineering Shenzhen Institute of Information Technology Guangdong China State Key Lab of Integrated Service Networks School of Cyber Engineering Xidian University Shaanxi China Cyberspace Institute of Advanced Technology Guangzhou University Guangdong China School of Computer Science University College Dublin Dublin Ireland

Noise in data appears to be inevitable in most realworld machine learning applications and would cause severe overfitting problems. Not only can data features contain noise, but labels are also prone to be noisy due to human input. In this paper, rather than noisy label learning in multiclass classifications, we instead focus on the less explored area of noisy label learning for multilabel classifications. Specifically, we investigate the post-correction of predictions generated from classifiers learned with noisy labels. The reasons are two-fold. Firstly, this approach can directly work with the trained models to save computational resources. Secondly, it could be applied on top of other noisy label correction techniques to achieve further improvements. To handle this problem, we appeal to deep generative approaches that are possible for uncertainty estimation. Our model posits that label noise arises from a stochastic shift in the latent variable, providing a more robust and beneficial means for noisy learning. We develop both unsupervised and semi-supervised learning methods for our model. The extensive empirical study presents solid evidence to that our approach is able to consistently improve the independent models and performs better than a number of existing methods across various noisy label settings. Moreover, a comprehensive empirical analysis of the proposed method is carried out to validate its robustness, including sensitivity analysis and an ablation study, among other elements. © 2025, CC BY.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Enhancing Group Privacy for Effective CIoT Data Mining in 6G Networks Using Differential Privacy

引用

IEEE Transactions on Consumer Electronics 2025年

作者： Ning, Bo Zhang, Xinjian Liu, Chengfei Wang, Ning Li, Guanyu Dalian Maritime University School of Information Science and Technology Liaoning Dalian116026 China Swinburne University of Technolgy Department of Computer Science and Software Engineering MelbourneVIC3122 Australia Guangzhou University Cyberspace Institute of Advanced Technology Guangdong Guangzhou511442 China

In the realm of 6G wireless networks, the Consumer Internet of Things (CIoT) aims to revolutionize consumer electronics by integrating advanced technologies such as Artificial Intelligence (AI). As CIoT environments become more dynamic and data-rich, ensuring data privacy while maintaining the utility of data mining is crucial. Graph data mining, which models object relationships with expressive capabilities, is central to this transformation but may pose data leakage risks. Traditional differential privacy methods in graphs often focus on releasing attribute or node information or adding excessive noise, reducing data availability. Therefore our focus is on protecting group relationships and leveraging their closeness to design budget allocation mechanisms that reduce noise. The proposed model uses k-cores to hierarchically manage group relationships, considering the variation in edges between k-cores and (k-1)-cores. Significant changes indicate a high degree of aggregation and interdependence at higher levels, warranting greater privacy protection. By allocating the privacy budget according to edge levels, our approach balances protection and data availability. Experiments on four types of graphs show that the network generated by our model accurately represents the original graph, making it suitable for subsequent data mining applications. © 2025 IEEE.

关键词： Information leakage

来源：评论

学校读者我要写书评

暂无评论

Empowering Deaf Children with Sinhala Sign Language, Emotion Detection, and Sound Recognition

Empowering Deaf Children with Sinhala Sign Language, Emotion...

引用

Advancements in Computing (ICAC), International Conference on

作者： W.H.K Fernando P.L.D.N Perera M.I.N Fernando P.T.E.W Gunasekara Harinda Fernando Rivoni De Zoysa Department of Information Technology Sri Lanka Institute of Information Technology Malabe Sri Lanka Department of Computer System Engineering Sri Lanka Institute of Information Technology Malabe Sri Lanka Department of Computer Science Software Engineering Sri Lanka Institute of Information Technology Malabe Sri Lanka

This study describes a broad endeavor to use cutting-edge technologies to empower deaf primary school students in Sri Lanka. Three key elements make up the study: a sound recognition and classification system, an Android software for translating gestures in Sinhala Sign Language (SSL), and a mobile app for emotion recognition and text-tospeech in Sinhala. By promptly alerting caregivers to potential dangers, the sound recognition technology secures the security of deaf children at home, providing an invaluable layer of security. The Android software bridges the communication gap and improves family interactions by translating SSL gestures captured by the device's camera into textual representations. This promotes successful communication between deaf children and their parents. Convolutional Neural Networks (CNNs) are used in the mobile application to detect and understand emotions based on facial expressions. This enables non-verbal youngsters to communicate successfully through Sinhala textto-speech output with adaptive tone modulation. This research aims to build a caring and inclusive atmosphere, enabling increased independence and enhanced communication skills in deaf children by combining these advances. The suggested solutions could significantly improve the lives of deaf kids, providing them with opportunities for better learning, social interaction, and emotional health.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multi-Modal Medical Vision-and-Language Learning for Retinal Vein Occlusion Classification

健康数据科学（英文）

引用

健康数据科学（英文） 2023年第4期 42-43页

作者： Weibin Liao Yanfeng Liao Zhimin Fan Jiexuan Zhang Shuochen Li Jiarui Yang Liantao Ma National Engineering Research Center for Software Engineering Peking UniversityBeijingChina Computer Science and Technology Beijing Institute of TechnologyBeijingChina Department of Ophthalmology Peking University Third HospitalBeijing Key Laboratory of Restoration of Damaged Ocular NervePeking University Third HospitalBeijingChina National Engineering Research Center for Software Engineering Peking UniversityBeijingChina

Background:Retinal vein occlusion(RVO)is one of the most common retinal vascular diseases leading to vision loss if not diagnosed and treated in *** diagnosis of central and branch RVO(CRVO and BRVO)can alleviate the workload of ophthalmologists while facilitating early detection and treat-ment of RVO and laying the foundation for subsequent symptom grading,treatment planning,and ***,the development of a fast,high-performance,and robust diagnostic model is a crucial step toward achieving quantitative and accu-rate assessment of RVO *** there is an extensive research using fundus visual images for clinical assisted diagno-sis with deep learning models,there is a lack of focus on incor-porating doctors'text reports to further enhance the models. Objectives:In this study,we propose a multi-modal medical visual-and-language learning model that utilizes fundus fluo-rescein angiography(FFA)images and physician analysis reports to classify BRVO and CRVO cases. Methods:As shown in Fig.1,the proposed model utilizes an advanced convolutional neural network to extract visual features from FFA images for patient visual representation *** language representation learning,the model first extracts basic patient features such as gender,age,vision,and blood pressure,and then uses regular expression matching to obtain typical patient symptoms from expert text ***,we cre-ated a sign vocabulary for RVO patients,including exudate,macular edema,and intraretinal hemorrhage,among *** on this vocabulary,the proposed model analyzes patient symptom manifestations and learns symptom presentation(such as exudate stage and location).Finally,based on the patient's visual and text representation,the proposed model uses a fully connected(FC)layer for classification tasks. Results:We evaluated the proposed model on a private dataset consisting of 101 patients and 1,265 FFA images,composed of 58 BRVO patients and 43 CRVO patie

关键词： language vision learning classification medical multi-modal occlusion retinal vein

来源：评论

学校读者我要写书评

暂无评论

Identification of Abnormal Behavior in Electric Power Operations Using the Human Skeleton's Key Features 8

Identification of Abnormal Behavior in Electric Power Operat...

引用

8th International Conference on Image, Vision and Computing, ICIVC 2023

作者： Zhou, Yang Wang, Xingjian Hu, Guoqiang Luo, Yong Chai, Zheng Xu, Wei Liu, Qiuming State Grid Jiangxi Electric Power Co Information and Communication Branch Nanchang330095 China Computer College Nanjing University of Posts and Telecomunications Nanjing210032 China State Grid Nanchang Power Supply Company Nanchang330095 China School of Software Jiangxi Normal University Nanchang330022 China School of Software Engineering Jiangxi University of Science and Technology Nanchang330013 China

ISBN: (纸本)9798350335231

At present, the safety monitoring of electric power sites mainly monitors the monitoring video through personnel, which has problems such as heavy workload, missed detection, and false alarms. In order to realize the intelligent recognition of the behavior of personnel at the electric power operation site, an abnormal behavior recognition method based on the key points of human skeleton is proposed, which can realize the real-time detection of the behavior of personnel at the operation site. First, the backbone network of YOLOv5s is improved for target detection, and the trajectory of the target is obtained by combining the DeepSort target tracking algorithm;then, the AlphaPose attitude estimation method is used to extract the target skeleton sequence from the trajectory;finally, the multi-partition spatiotemporal graph convolutional network is used to track the personnel Abnormal behavior identification. The experimental results show that the recognition accuracy rate of the algorithm reaches 93.5%, and the processing speed reaches 18.32 fps, which can identify abnormal behaviors of electric power workers in real time and accurately. © 2023 IEEE.

关键词： Personnel

来源：评论

学校读者我要写书评

暂无评论

Improving Local Search for Pseudo Boolean Optimization by Fragile Scoring Function and Deep Optimization 29

Improving Local Search for Pseudo Boolean Optimization by Fr...

引用

29th International Conference on Principles and Practice of Constraint Programming, CP 2023

作者： Zhou, Wenbo Zhao, Yujiao Wang, Yiyuan Cai, Shaowei Wang, Shimao Wang, Xinyu Yin, Minghao School of Information Science and Technology Northeast Normal University Changchun China Key Laboratory of Applied Statistics of MOE Northeast Normal University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering of MOE Jilin University Changchun 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 Beijing China

ISBN: (纸本)9783959773003

Pseudo-Boolean optimization (PBO) is usually used to model combinatorial optimization problems, especially for some real-world applications. Despite its significant importance in both theory and applications, there are few works on using local search to solve PBO. This paper develops a novel local search framework for PBO, which has three main ideas. First, we design a two-level selection strategy to evaluate all candidate variables. Second, we propose a novel deep optimization strategy to disturb some search spaces. Third, a sampling flipping method is applied to help the algorithm jump out of local optimum. Experimental results show that the proposed algorithms outperform three state-of-the-art PBO algorithms on most instances. © Wenbo Zhou, Yujiao Zhao, Yiyuan Wang, Shaowei Cai, Shimao Wang, Xinyu Wang, and Minghao Yin.

关键词： Combinatorial optimization

来源：评论

学校读者我要写书评

暂无评论

Digital Assistant for Visually Impaired People

Digital Assistant for Visually Impaired People

引用

Advancements in Computing (ICAC), International Conference on

作者： Thuvaraka Thayaparan Sanjay Jayakumar B. A. I Fernando Gobisan Ananthanadarajan Lakmini Abeywardana Dharshana Kasthurirathna Department of Information Technology Sri Lanka Institute of Information Technology Malabe Sri Lanka Computer Science and Software Engineering Sri Lanka Institute of Information Technology Malabe Sri Lanka

This research paper presents a comprehensive exploration of various components to enhance the capabilities of a digital assistant tailored for visually impaired individuals. The first component explores how various image captioning architectures and Vision Encoder Decoder models can be used for better environmental interaction. The second component concerns with facial expression recognition for better social interaction and the third component is a currency note identification component that aims to aid visually impaired individuals with cash transaction whilst the final component that explores a voice bot aims to improve a voice bot and how to manage accents, speech accuracy and naturality using modern deep learning techniques. By investigating these components, this research advances assistive technology, empowering visually impaired individuals with a conversational digital assistant that enhances environmental interaction, supports social interactions, and aids in currency identification.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SWPFOPLD: A Profiling and Optimizing Loader for SW26010Pro Processors

SWPFOPLD: A Profiling and Optimizing Loader for SW26010Pro P...

引用

International Conference on computers, Communications, and Systems (ICCCS)

作者： Hong Qian Wei Wu Qi Zhu Fei Wang Jinwei Zhao Yan Zheng School of Computer Science and Technology University of Science and Technology of China Hefei China System Software Research Department Jiangnan Institute of Computing Technology Wuxi China Department of Computer Science and Technology Tsinghua University Beijing China National Research Center of Parallel Computer Engineering and Technology Beijing China

The Sunway family supercomputers have achieved a series of remarkable achievements. However, the toolchains provided by them are not perfect, which has brought great challenges to the development of high-performance application software. In this paper, a profiling and optimizing tool is proposed to assist people to analyze and optimize the performance of their programs. SWPFOPLD is independent of the application program and gathers the runtime performance data through the PMUs, an automatically hot functions rearrange optimization based on the performance data is furtherly accomplished. The evaluation shows that SWPFOPLD can be easily and effectively used to analyze and optimize the performance of the application programs.

关键词：

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：