检索结果-内蒙古大学图书馆

Research of Malicious Code Attack Effect Based on Synthetic Entropy Method

Chinese Journal of Electronics 2023年第3期22卷 449-454页

作者： LI Peng WANG Ruchuan College of Computer Nanjing University of Posts and Telecommunications Nanjing China Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks Nanjing China Key Lab of Broadband Wireless Communication and Sensor Network Technology Nanjing University of Posts and Telecommunications Ministry of Education Jiangsu Province Nanjing China

To resolve the problem of assessing the effect of a network attack, this paper combines cross entropy with network character entropy method and proposes scheme to evaluate the malicious code attack effect. It captures the related indicators in real time and normalizes the data so as to evaluate it at the same level. It adopts cross entropy method to pretreat the indicators adaptively. We calculate the weight coefficient and exploit network character entropy method to evaluate the attack with accuracy according to the importance of the indicators in the evaluation system. Experimental results and corresponding comparisons reveal that the proposed method can quantitatively determine the exact effect of the malicious code attack.

关键词： Industries Medical services Entropy Malware Real-time systems Information technology

来源：评论

学校读者我要写书评

暂无评论

Membership Inference Attacks via Dynamic Adversarial Perturbations Reduction

Membership Inference Attacks via Dynamic Adversarial Perturb...

引用

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Zehua Ding Youliang Tian Guorong Wang Jinbo Xiong Jianfeng Ma State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang China Fujian Provincial Key Laboratory of Network Security and Cryptology College of Computer and Cyber Security Fujian Normal University Fuzhou China School of Cyber Engineering Xidian University Xian China

ISBN: (数字)9798331506209

ISBN: (纸本)9798331506216

Existing membership inference attacks (MIAs) based on adversarial attacks typically introduce excessively large adversarial perturbations to change the model predictions. However, such methods are not only likely to obscure the characteristic memory of the model regarding member data but also blur the subtle characteristic differences between member and non-member data. To address this issue, we propose a novel dynamic adversarial perturbation reduction MIA (DAPR-MIA), which aims to enhance the ability to identify the membership of samples by reducing the perturbation strength. Specifically, DAPR-MIA first conducts a fine-grained analysis of the gradient components of conventional adversarial examples, assessing the impact of each element on the model decision to generate a sensitivity mask. This mask identifies dimensions within adversarial examples that significantly influence model predictions, referred to as sensitive dimensions. Then, by dynamically reducing the perturbation size in these sensitive dimensions, DAPR-MIA gradually approaches the minimum perturbation required to change the prediction result of samples, maximizing the retention of membership features within the sample. Finally, the deviation of the prediction results between the original sample and the adversarial example after perturbation attenuation on the target model is measured to infer whether the sample belongs to a member data. Experimental results show that DAPR-MIA significantly outperforms existing methods on multiple datasets and deep models, demonstrating state-of-the-art attack accuracy.

关键词： Privacy Sensitivity Perturbation methods Computational modeling Predictive models Attenuation measurement Attenuation Data models Robustness Security

来源：评论

学校读者我要写书评

暂无评论

Unraveling Intricacies: A Decomposition Approach for Few-Shot Multi-Intent Spoken Language Understanding

Unraveling Intricacies: A Decomposition Approach for Few-Sho...

引用

2024 IEEE International Conference on Big Data, BigData 2024

作者： Hua, Wenbin Wang, Yufan Fan, Rui Tu, Xinhui He, Tingting Central China Normal University Hubei Provincial Key Laboratory of Artificial Intelligence and Smart Learning National Language Resources Monitor Research Center for Network Media Wuhan China Central China Normal University School of Computer Wuhan China Shenzhen Technology University College of Big Data and Internet Shenzhen China Central China Normal University Faculty of Artificial Intelligence in Education Wuhan China

ISBN: (纸本)9798350362480

Few-shot multi-intent spoken language understanding (SLU) aims to detect user's multiple intents and key slots using a tiny amount of annotated data. Prevailing multi-intent SLU models typically rely on abundant data for effective training, enabling them to capture corresponding relationships between intents and slots. However, in few-shot scenarios, establishing these connections becomes challenging, especially in situations involving multiple intents, which may result in confused relationships between intents and slots. To overcome the challenge, we propose decomposing the multi-label intent detection task into several single-label tasks, which reduces the complexity of model training while preserving the constraint relationships between intents and their related slots. We design description templates for each intent and respectively predict the correlation between the utterance and each intent description while completing the slot filling task under the corresponding intent. Therefore, each intent can independently guide the slot filling process, mitigating potentially confused relationships between multiple intents and slots. Experimental results on public datasets indicate that the performance of our model is better than ChatGPT and achieves state-of-the-art results. © 2024 IEEE.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

CHECKWHY: Causal Fact Verification via Argument Structure

arXiv

引用

arXiv 2024年

作者： Si, Jiasheng Zhao, Yibo Zhu, Yingjie Zhu, Haiyang Lu, Wenpeng Zhou, Deyu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Qilu University of Technology Shandong Academy of Sciences China School of Computer Science and Engineering Southeast University China Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications Southeast University Ministry of Education China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science China

With the growing complexity of fact verification tasks, the concern with "thoughtful" reasoning capabilities is increasing. However, recent fact verification benchmarks mainly focus on checking a narrow scope of semantic factoids within claims and lack an explicit logical reasoning process. In this paper, we introduce CHECKWHY, a challenging dataset tailored to a novel causal fact verification task: checking the truthfulness of the causal relation within claims through rigorous reasoning steps. CHECKWHY consists of over 19K "why" claim-evidence-argument structure triplets with supports, refutes, and not enough info labels. Each argument structure is composed of connected evidence, representing the reasoning process that begins with foundational evidence and progresses toward claim establishment. Through extensive experiments on state-of-the-art models, we validate the importance of incorporating the argument structure for causal fact verification. Moreover, the automated and human evaluation of argument structure generation reveals the difficulty in producing satisfying argument structure by fine-tuned models or Chain-of-Thought prompted LLMs, leaving considerable room for future improvements1 Copyright © 2024, The Authors. All rights reserved.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Membership Inference Attacks via Dynamic Adversarial Perturbations Reduction 23

Membership Inference Attacks via Dynamic Adversarial Perturb...

引用

23rd IEEE International Conference on Trust, Security and Privacy in Computing and Communications, TrustCom 2024

作者： Ding, Zehua Tian, Youliang Wang, Guorong Xiong, Jinbo Ma, Jianfeng Guizhou University State Key Laboratory of Public Big Data College of Computer Science and Technology Guiyang China Fujian Normal University Fujian Provincial Key Laboratory of Network Security and Cryptology College of Computer and Cyber Security Fuzhou China Xidian University School of Cyber Engineering Xian China

关键词： Perturbation techniques

来源：评论

学校读者我要写书评

暂无评论

A Multi-Channel Composite Recommendation Method in Federated Learning

A Multi-Channel Composite Recommendation Method in Federated...

引用

2023 IEEE International Conference on Dependable, Autonomic and Secure Computing, 2023 International Conference on Pervasive Intelligence and Computing, 2023 International Conference on Cloud and Big Data Computing, 2023 International Conference on Cyber Science and technology Congress, DASC/PiCom/CBDCom/CyberSciTech 2023

作者： Chen, Shi Li, Qianmu Wang, Qi Zhao, Xinjian Ge, Hao Li, Nianzhe State Grid Jiangsu Information & Telecommunication Company Nanjing China School of Computer Science Ana Engineering Nanjing University of Science and Technology Nanjing China State Grid Smart Grid Research Institute Co. Ltd. State Grid Key Laboratory of Information & Network Security Nanjing China

ISBN: (纸本)9798350304602

There are various data security risks in power grid systems, and recommendation systems are a good solution to the selection of security strategies. Federated learning is a machine learning framework that can protect the data privacy of each participant and can effectively solve the data security external collaborative training modeling challenges. This paper proposed a flexible implantable recommendation algorithm based on knowledge graph and multi-channel composite to enhance federated learning on demand. The algorithm designs three different feature extraction modules to learn to construct multi-dimensional features on multi-attribute collaborative knowledge graphs. Among them, the distance translation module utilizes the Trans R recommendation algorithm to learn the internal semantic structure information in knowledge graph triples and terminal-security policy selection history;the random walk module uses the Node2Vec algorithm to learn the homogeneous and Structured information in the multi-attribute collaborative knowledge graph;the graph convolutional network module learns the high-order neighborhood multi-attribute collaborative information of terminal/security policy nodes through multi-layer feature propagation and aggregation. The composite network module adopts a convolutional neural network-like operation to first locally optimize the three modules, and then extracts the highly informative elements to obtain the final feature representation of the terminal/security policy. Finally, the algorithm merges the vectors of terminal-security policy pairs into the multi-layer perceptron to learn and calculate the prediction score for recommendation. © 2023 IEEE.

关键词： Recommender systems

来源：评论

学校读者我要写书评

暂无评论

NextSeg: Automatic COVID-19 Lung Infection Segmentation from CT Images Based on Next-ViT

NextSeg: Automatic COVID-19 Lung Infection Segmentation from...

引用

International Joint Conference on Neural networks (IJCNN)

作者： Aimei Dong Ruixin Wang Xuening Zhang Jian Liu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Faculty of Computer Science and Technology Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

The spread of COVID-19 has brought great disaster to the world, and automatic segmentation of the infected region can help doctors make a quick diagnosis and reduce workload. However, automatic segmentation of infected regions in computed tomography (CT) images still faces many challenges. The most central problem at the moment is that when capturing low-frequency signals and high-frequency signals, the use of Transformer blocks for capturing low-frequency signals may deteriorate the high-frequency signals (local texture information) to some extent. For this reason, in this paper, we propose a new network for automatic segmentation of COVID-19 lung infections from CT images, called NextSeg. NextSeg follows an encoder-decoder architecture, where the encoder part uses NCB and NTB convolutional attention modules to extract better features while reducing the deterioration of texture profile information. The decoder part uses an attention fusion module to learn the weight relationship between features and retain more important feature information. In addition to the difficulty of extracting the boundary details of the COVID-19 infected region as well as the loss of details the deeper the network is, we designed a new hybrid strategy. Traditional and state-of-the-art approaches to the COVID-19 segmentation domain are compared. Extensive experiments show that our proposed framework outperforms existing competitors in both qualitative and quantitative results.

关键词： COVID-19 Image segmentation Convolution Computed tomography Neural networks Lung Medical services

来源：评论

学校读者我要写书评

暂无评论

Dbfl: Dynamic Byzantine-Robust Privacy Preserving Federated Learning in Heterogeneous Data Scenario

SSRN

引用

SSRN 2024年

作者： Chen, Xiaoli Tian, Youliang Wang, Shuai Yang, Kedi Zhao, Wei Xiong, Jinbo State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guizhou Guiyang550025 China Guizhou Provincial Key Laboratory of Cryptography and Blockchain Technology Guizhou University Guizhou Guiyang550025 China College of Mathematics and Statistics Guizhou University Guizhou Guiyang550025 China Fujian Provincial Key Laboratory of Network Security and Cryptology College of Computer and Cyber Security Fujian Normal University Fujian Fuzhou350117 China

Privacy Preserving Federated Learning (PPFL) protects the clients' local data privacy by securely uploading encrypted gradients to the server. However, in real-world scenarios, heterogeneous client data distribution makes identifying poisoning gradients challenging. And during local iterations, models moving in different directions constantly shift the boundary between benign and malicious gradients. To address these challenges, we design a Dynamic Byzantine-robust defense strategy (DBFL) based on two-trapdoor homomorphic encryption, achieving detection of encrypted model poisoning attacks. Specifically, we first propose a secure Manhattan Distance method, which accurately measures the difference between each element in two encrypted gradients, thus enabling precise detection of poisoning attacks in heterogeneous data scenarios without exposing any privacy. Furthermore, we design a Byzantine-tolerant aggregation mechanism based on dynamic threshold, where the dynamic threshold changes according to the continuously changing boundary between poisoning and benign gradients in heterogeneous data scenarios. Even when malicious clients controlled by Byzantine attackers constitute 70\% of the DBFL, our method accurately excludes poisoned gradients. Security analysis demonstrates that DBFL achieves IND-CPA security. Extensive evaluation on two benchmark datasets shows that DBFL outperforms existing defense strategies. In particular, DBFL achieves a 15\%-40\% accuracy improvement in the non-IID setting compared to existing solutions. © 2024, The Authors. All rights reserved.

关键词： network security

来源：评论

学校读者我要写书评

暂无评论

A Dirty Data Identification Method in Mechanical Equipment Condition Monitoring

A Dirty Data Identification Method in Mechanical Equipment C...

引用

Smart World Congress (SWC), IEEE

作者： Huijuan Hao Hongge Zhao Sijian Zhu Yue Feng Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China Jinan China

ISBN: (数字)9798331520861

ISBN: (纸本)9798331520878

The rise of the Industrial Internet of Things (IIOT) has brought new opportunities for the mechanical equipment condition monitoring. The trend towards intelligent and automated production of mechanical equipment has become mainstream. However, in the actual operational environment, mechanical equipment is often mixed with a large amount of dirty data unrelated to equipment failures, which seriously affects equipment fault diagnosis and wastes maintenance resources. To improve the reliability of mechanical equipment health status analysis, this paper proposes a new method for dirty data identification suitable for mechanical equipment condition monitoring. First, utilizing the time-space position encoding (TSPE) to map positional relationships between sequences. Next, designing a time-dimension attention (TDA) mechanism to capture the correlation between multiple temporal attribute features and the temporal dependencies between preceding and subsequent data. Then, building a dirty data identification model using an improved Transformer network and Generative Adversarial network (GAN). Finally, employing an adaptive thresholding method to dynamically set thresholds, determining the degree of data anomalies by comparing anomaly scores with the threshold. The effectiveness of the proposed method are validated through one simulation experiment and one real-world example. The F1 Score of this method on the simulated missing data and drifting data is 0.9874 and 0.9838. On the drive-end and fan-end is $\mathbf{0. 9 7 0 0}$ and 0.9882 at Case Western Reserve University (CWRU). Compared to other dirty data identification methods in the field, this method has better dirty data identification capability.

关键词： Condition monitoring Correlation Buildings Production Transformers Generative adversarial networks Market research Data models Monitoring Industrial Internet of Things

来源：评论

学校读者我要写书评

暂无评论

An Approach to New Bearing Fault Diagnosis Utilizing Deep Transfer Learning

An Approach to New Bearing Fault Diagnosis Utilizing Deep Tr...

引用

Smart World Congress (SWC), IEEE

作者： HuiJuan Hao Yue Feng HongGe Zhao SiJian Zhu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Industrial Network and Information System Security Shandong Fundamental Research Center for Computer Science Jinan China Jinan China

ISBN: (数字)9798331520861

ISBN: (纸本)9798331520878

Bearings, serving as crucial components in mechanical equipment, exert a direct influence on the overall performance of the system. Deep transfer learning has garnered significant attention in the realm of intelligent fault diagnosis for machines. Nevertheless, extant deep transfer learning algorithms encounter challenges when detecting newly emerging faults within the target domain. To address this limitation, this study introduces a novel bearing fault diagnosis approach for newly emerging faults, leveraging deep transfer learning principles. Initially, a multi-grained hierarchical feature extraction network is constructed to effectively capture intricate data characteristics. Subsequently, these extracted features are employed to train a fault classifier, thereby establishing a fault diagnosis network grounded in deep transfer learning. This network is capable of discriminating unlabeled new faults from labeled known faults with remarkable precision. Furthermore, an adaptive hierarchical clustering algorithm, informed by the elbow method, is incorporated to categorize the newly emerging faults. The clustered samples are then utilized to train the online sequential extreme learning machine (OS-ELM), facilitating the construction of a new fault diagnosis network tailored for diagnosing these faults in the target domain. The efficacy of the proposed methodology was rigorously verified through experiments executed on both the CWRU and XJTU-SY datasets. The outcomes conclusively demonstrated an accuracy of $\mathbf{9 8. 7 5 \%}$ on the CWRU dataset and $\mathbf{9 7. 9 2 \%}$ on the XJTU-SY dataset. This model portends immense potential in tackling fault diagnosis transfer scenarios characterized by the emergence of multiple novel faults within the target domain, thereby markedly broadening the applicability of deep transfer learning.

关键词： Fault diagnosis Accuracy Adaptive systems Extreme learning machines Transfer learning Clustering algorithms Feature extraction Classification algorithms Data mining Elbow

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：