检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Zhao, Jian Wang, Shenao Zhao, Yanjie Hou, Xinyi Wang, Kailong Gao, Peiming Zhang, Yuanchao Wei, Chen Wang, Haoyu Huazhong University of Science and Technology Wuhan China MYbank Ant Group Hangzhou China Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology China

The proliferation of pre-trained models (PTMs) and datasets has led to the emergence of centralized model hubs like Hugging Face, which facilitate collaborative development and reuse. However, recent security reports have uncovered vulnerabilities and instances of malicious attacks within these platforms, highlighting growing security concerns. This paper presents the first systematic study of malicious code poisoning attacks on pre-trained model hubs, focusing on the Hugging Face platform. We conduct a comprehensive threat analysis, develop a taxonomy of model formats, and perform root cause analysis of vulnerable formats. While existing tools like Fickling and ModelScan offer some protection, they face limitations in semantic-level analysis and comprehensive threat detection. To address these challenges, we propose MalHug, an end-to-end pipeline tailored for Hugging Face that combines dataset loading script extraction, model deserialization, in-depth taint analysis, and heuristic pattern matching to detect and classify malicious code poisoning attacks in datasets and models. In collaboration with Ant Group, a leading financial technology company, we have implemented and deployed MalHug on a mirrored Hugging Face instance within their infrastructure, where it has been operational for over three months. During this period, MalHug has monitored more than 705K models and 176K datasets, uncovering 91 malicious models and 9 malicious dataset loading scripts. These findings reveal a range of security threats, including reverse shell, browser credential theft, and system reconnaissance. This work not only bridges a critical gap in understanding the security of the PTM supply chain but also provides a practical, industry-tested solution for enhancing the security of pre-trained model hubs. Copyright © 2024, The Authors. All rights reserved.

关键词： Latent semantic analysis

来源：评论

学校读者我要写书评

暂无评论

Breaking Barriers in Physical-World Adversarial Examples: Improving Robustness and Transferability via Robust Feature

arXiv

引用

arXiv 2024年

作者： Wang, Yichen Chou, Yuxuan Zhou, Ziqi Zhang, Hangtao Wan, Wei Hu, Shengshan Li, Minghui National Engineering Research Center for Big Data Technology and System China Services Computing Technology and System Lab China Cluster and Grid Computing Lab China Hubei Engineering Research Center on Big Data Security China Hubei Key Laboratory of Distributed System Security China School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Computer Science and Technology Huazhong University of Science and Technology China School of School of Software Engineering Huazhong University of Science and Technology China

As deep neural networks (DNNs) are widely applied in the physical world, many researches are focusing on physical-world adversarial examples (PAEs), which introduce perturbations to inputs and cause the model’s incorrect outputs. However, existing PAEs face two challenges: unsatisfactory attack performance (i.e., poor transferability and insufficient robustness to environment conditions), and difficulty in balancing attack effectiveness with stealthiness, where better attack effectiveness often makes PAEs more perceptible. In this paper, we explore a novel perturbation-based method to overcome the challenges. For the first challenge, we introduce a strategy Deceptive RF injection based on robust features (RFs) that are predictive, robust to perturbations, and consistent across different models. Specifically, it improves the transferability and robustness of PAEs by covering RFs of other classes onto the predictive features in clean images. For the second challenge, we introduce another strategy Adversarial Semantic Pattern Minimization, which removes most perturbations and retains only essential adversarial patterns in AEs. Based on the two strategies, we design our method Robust Feature Coverage Attack (RFCoA), comprising Robust Feature Disentanglement and Adversarial Feature Fusion. In the first stage, we extract target class RFs in feature space. In the second stage, we use attention-based feature fusion to overlay these RFs onto predictive features of clean images and remove unnecessary perturbations. Experiments show our method’s superior transferability, robustness, and stealthiness compared to existing state-of-the-art methods. Additionally, our method’s effectiveness can extend to Large Vision-Language Models (LVLMs), indicating its potential applicability to more complex tasks. Copyright © 2024, The Authors. 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

Layered Structure Aware Dependent Microservice Placement Toward Cost Efficient Edge Clouds 42

Layered Structure Aware Dependent Microservice Placement Tow...

引用

42nd IEEE International Conference on Computer Communications, INFOCOM 2023

作者： Zeng, Deze Geng, Hongmin Gu, Lin Li, Zhexiong China University of Geosciences School of Computer Science Wuhan China Huazhong University of Science and Technology National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Wuhan China

ISBN: (纸本)9798350334142

Although the containers are featured by light-weightness, it is still resource-consuming to pull and startup a large container image, especially in relatively resource-constrained edge cloud. Fortunately, Docker, as the most widely used container, provides a unique layered architecture that allows the same layer to be shared between microservices so as to lower the deployment cost. Meanwhile, it is highly desirable to deploy dependent microservices of an application together to lower the operation cost. Therefore, the balancing of microservice deployment cost and the operation cost should be considered comprehensively to achieve minimal overall cost of an on-demand application. In this paper, we first formulate this problem into a Quadratic Integer Programming form (QIP) and prove it as a NP-hard problem. We further propose a Randomized Rounding-based Microservice Deployment and Layer Pulling (RR-MDLP) algorithm with low computation complexity and guaranteed approximation ratio. Through extensive experiments, we verify the high efficiency of our algorithm by the fact that it significantly outperforms existing state-of-the-art microservice deployment strategies. © 2023 IEEE.

关键词： Containers

来源：评论

学校读者我要写书评

暂无评论

Intelligent Smishing Detection System for Massive Users 9

Intelligent Smishing Detection System for Massive Users

引用

9th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2023

作者： Li, Qiang Zhang, Jianxin Zhou, Nan Ming, Liang Zhao, Bo Du, Yuejin Zhong, Li Li, Jia Beijing Qihoo Technology Co. Ltd Beijing100015 China National Engineering Research Center of Big Data Collaborative Security Technology Beijing100015 China WuHan University School of Cyber Science and Engineering WuHan430064 China

ISBN: (纸本)9798350304428

Smishing has always been an important means of black-gray drainage. Currently, interception measures are primarily deployed at the network communication layer of SMS transmission. In response, black-gray industry organizations continuously improve their evasion techniques to bypass the interception strategies along the communication chain. This article presents a novel phishing text message interception method that involves collaboration between the terminal and the cloud. Its advantage lies in the combination of deep learning models for text classification and the identification of reserved information. Firstly, the optimization process involves leveraging a massive corpus in the vertical domain, consisting of hundreds of millions of samples, to optimize the Bert model and TextCNN network. This optimization includes incorporating proprietary features such as statistics, pinyin, and glyph. As a result, classification models are specifically designed for both the cloud and terminals, enabling efficient recognition of ordinary phishing text messages. Secondly, to address spear-phishing attacks and various types of spoofing in phishing text messages, an enhanced named entity recognition technology is employed. This technology allows for the extraction of URLs, QQ numbers, WeChat IDs, mobile phone numbers, and other suspected fraudulent reserved information from the text messages. Subsequently, dedicated models for feature fusion of URLs and heuristic discrimination strategies for network accounts are established based on this extracted information. Combining the two aforementioned detection methods, the paper has implemented an intelligent phishing text message detection system, which has been deployed and applied in a real environment with of hundred millions users. The system achieved a recall rate of 98.32% and an accuracy rate of 99.44%. © 2023 IEEE.

关键词： Character recognition

来源：评论

学校读者我要写书评

暂无评论

Resource abstraction and data placement for distributed hybrid memory pool

引用

Frontiers of Computer science 2021年第3期15卷 47-57页

作者： Tingting CHEN Haikun LIU Xiaofei LIAO Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab/Gluster and Grid Computing LabSchool of Gomputing Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

Emerging byte-addressable non-volatile memory(NVM)technologies offer higher density and lower cost than DRAM,at the expense of lower performance and limited write *** have been many studies on hybrid NVM/DRAM memory management in a single physical ***,it is still an open problem on how to manage hybrid memories efficiently in a distributed *** paper proposes Alloy,a memory resource abstraction and data placement strategy for an RDMA-enabled distributed hybrid memory pool(DHMP).Alloy provides simple APIs for applications to utilize DRAM or NVM resource in the DHMP,without being aware of the hardware details of the *** propose a hotness-aware data placement scheme,which combines hot data migration,data replication and write merging together to improve application performance and reduce the cost of *** evaluate Alloy with several micro-benchmark workloads and public benchmark *** results show that Alloy can significantly reduce the DRAM usage in the DHMP by up to 95%,while reducing the total memory access time by up to 57%compared with the state-of-the-art approaches.

关键词： load balance distributed hybrid memory clouds

来源：评论

学校读者我要写书评

暂无评论

Privacy Law Enforcement Under Centralized Governance: A Qualitative Analysis of Four Years’ Special Privacy Rectification Campaigns

arXiv

引用

arXiv 2025年

作者： Jing, Tao Li, Yao Ye, Jingzhou Wang, Jie Wang, Xueqiang School of Cyber Science and Engineering Huazhong University of Science and Technology China JinYinHu Laboratory China University of Central Florida United States Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology China

In recent years, major privacy laws like the GDPR have brought about positive changes. However, challenges remain in enforcing the laws, particularly due to under-resourced regulators facing a large number of potential privacy-violating software applications (apps) and the high costs of investigating them. Since 2019, China has launched a series of privacy enforcement campaigns known as Special Privacy Rectification Campaigns (SPRCs) to address widespread privacy violations in its mobile application (app) ecosystem. Unlike the enforcement of the GDPR, SPRCs are characterized by large-scale privacy reviews and strict sanctions, under the strong control of central authorities. In SPRCs, central government authorities issue administrative orders to mobilize various resources for market-wide privacy reviews of mobile apps. They enforce strict sanctions by requiring privacy-violating apps to rectify issues within a short timeframe or face removal from app stores. While there are a few reports on SPRCs, the effectiveness and potential problems of this campaign-style privacy enforcement approach remain unclear to the community. In this study, we conducted 18 semi-structured interviews with app-related engineers involved in SPRCs to better understand the campaign-style privacy enforcement. Based on the interviews, we reported our findings on a variety of aspects of SPRCs, such as the processes that app engineers regularly follow to achieve privacy compliance in SPRCs, the challenges they encounter, the solutions they adopt to address these challenges, and the impacts of SPRCs, etc. We found that app engineers face a series of challenges in achieving privacy compliance in their apps. For example, they receive inconsistent app privacy review reports from multiple app stores and have difficulties confirming the issues flagged by these reports;they also lack institutional support for studying privacy laws, self-validating privacy compliance of their apps, communicating effectiv

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

CLNX: Bridging Code and Natural Language for C/C++ Vulnerability-Contributing Commits Identification 39

CLNX: Bridging Code and Natural Language for C/C++ Vulnerabi...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Qin, Zeqing Wu, Yiwei Han, Lansheng School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan China Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security China Wuhan JinYinHu Laboratory China

ISBN: (纸本)157735897X

Large Language Models (LLMs) have shown great promise in vulnerability identification. As C/C++ comprise half of the open-source Software (OSS) vulnerabilities over the past decade and updates in OSS mainly occur through commits, enhancing LLMs' ability to identify C/C++ Vulnerability-Contributing Commits (VCCs) is essential. However, current studies primarily focus on further pre-training LLMs on massive code datasets, which is resource-intensive and poses efficiency challenges. In this paper, we enhance the ability of BERT-based LLMs to identify C/C++ VCCs in a lightweight manner. We propose CodeLinguaNexus (CLNX) as a bridge facilitating communication between C/C++ programs and LLMs. Based on commits, CLNX efficiently converts the source code into a more natural representation while preserving key details. Specifically, CLNX first applies Structure-level Naturalization to decompose complex programs, followed by Token-level Naturalization to interpret complex symbols. We evaluate CLNX on public datasets of 25,872 C/C++ functions with their commits. The results demonstrate that CLNX substantially improves the ability of LLMs to detect C/C++ VCCs. Moreover, CLNX-equipped CodeBERT achieves new state-of-the-art performance and identifies 38 OSS vulnerabilities in the real world. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Open source software

来源：评论

学校读者我要写书评

暂无评论

Large-Scale Network Adaptive Situation Awareness Method in Spatio-Temporal Dimension 1

引用

1st International Artificial Intelligence Conference, IAIC 2023

作者： Zhang, Hongbin Xu, Ying Liu, Bin Zhao, Dongmei Bai, Yikang School of Information Science and Engineering Hebei University of Science and Technology Shijiazhuang050000 China Hebei Key Laboratory of Network and Information Security Hebei Normal University Hebei Shijiazhuang050024 China School of Economics and Management Hebei University of Science and Technology Shijiazhuang050000 China Research Center of Big Data and Social Computing Hebei University of Science Shijiazhuang China

ISBN: (数字)9789819713325

ISBN: (纸本)9789819713318

In large-scale networks, the state space is exploding and changing dynamically. This leads to difficulties in collecting and analyzing situational awareness data, so we construct an adaptive situational awareness model in spatio-temporal dimensions. In the spatial dimension, vulnerabilities’s threats are assessed through attack graphs combined with Shapley values. At the same time, vulnerability threats are dynamically quantified by updating the status node reachability probability in real time. In the temporal dimension, a game model is established by analyzing vulnerability attack graph nodes to dynamically adjust the observation frequency of high-risk vulnerabilities, focusing on the safety status characteristics of high-risk assets. Experimental results show that our proposal integrates the security features of both space and time dimensions. This method can better focus on high-risk vulnerabilities and accurately reflect the dynamic changes in the network security situation, ensuring timeliness and accuracy in network security detection. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Network security

来源：评论

学校读者我要写书评

暂无评论

BadToken: Token-level Backdoor Attacks to Multi-modal Large Language Models

arXiv

引用

arXiv 2025年

作者： Yuan, Zenghui Shi, Jiawen Zhou, Pan Gong, Neil Zhenqiang Sun, Lichao Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology China Duke University United States Lehigh University United States

Multi-modal large language models (MLLMs) extend large language models (LLMs) to process multi-modal information, enabling them to generate responses to image-text inputs. MLLMs have been incorporated into diverse multimodal applications, such as autonomous driving and medical diagnosis, via plug-and-play without fine-tuning. This deployment paradigm increases the vulnerability of MLLMs to backdoor attacks. However, existing backdoor attacks against MLLMs achieve limited effectiveness and stealthiness. In this work, we propose BadToken, the first token-level backdoor attack to MLLMs. BadToken introduces two novel backdoor behaviors: Token-substitution and Token-addition, which enable flexible and stealthy attacks by making token-level modifications to the original output for backdoored inputs. We formulate a general optimization problem that considers the two backdoor behaviors to maximize the attack effectiveness. We evaluate BadToken on two open-source MLLMs and various tasks. Our results show that our attack maintains the model’s utility while achieving high attack success rates and stealthiness. We also show the real-world threats of BadToken in two scenarios, i.e., autonomous driving and medical diagnosis. Furthermore, we consider defenses including fine-tuning and input purification. Our results highlight the threat of our attack. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Optimized Dynamic Watermarking for Audio DNNs with Adaptive Embedding and Boundary Sampling

Optimized Dynamic Watermarking for Audio DNNs with Adaptive ...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Hao Fei Hewang Nie Siqi Sun Songfeng Lu Ting Luo Ling Qian Dunbo Cai Zhiguo Huang Runqing Zhang Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan China Shenzhen Huazhong University of Science and Technology Research Institute Shenzhen China China Mobile (Suzhou) Software Technology Company Limited Suzhou China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

The intensified concerns arising from the widespread adoption of deep learning have led to increased scrutiny of intellectual property protection in DNN models. Existing audio watermarking techniques, predominantly based on traditional signal processing methods, struggle to balance robustness, imperceptibility, and defense resistance in the face of evolving adversarial attacks. These limitations underscore the urgent need for more effective watermarking solutions in the audio domain. In this paper, we propose a dynamic audio watermarking framework that introduces an optimization-based approach to attach robust and adaptable triggers at arbitrary positions within audio signals, and innovatively integrates boundary sample selection driven by forgetting events and an adaptive watermark trigger embedding technique based on the SNR. Comprehensive experimental results reveal that our scheme preserves high model performance while maintaining remarkable stealthiness and robustness, offering a secure and reliable solution for safeguarding intellectual property in the audio domain and advancing the field of DNN watermarking.

关键词： Resistance Deep learning Adaptation models Watermarking Intellectual property Signal processing Robustness Speech processing Protection Signal to noise ratio

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：