检索结果-内蒙古大学图书馆

3rd International Conference on Machine Learning, Cloud Computing and Intelligent Mining, MLCCIM 2024

作者： Wu, Zhendong Li, Hu Zhang, Ming Lu, Shuaibing National Key Laboratory of Science and Technology on Information System Security Beijing China

ISBN: (纸本)9789819624676

Deep learning models are widely used in security-sensitive tasks such as facial recognition and autonomous driving. security issues in deep models could have serious implications for people’s lives, such as life-threatening situations resulting from autonomous driving failures. To enhance the security of deep models, security testing is necessary before their formal deployment. Since different scenarios may involve different deep models, a security testing method that targets multiple scenarios would be more universally applicable. In this paper, we propose a security testing method for deep models that can be applied across multiple scenarios, primarily utilizing adversarial attacks to test the models. For the deep model under test, we first construct threat scenarios and then associate the model with datasets and adversarial attack methods. Subsequently, we generate adversarial examples to input into the deep model to check its accuracy in classification. Finally, we evaluate the success rate of adversarial attacks on the deep model. A high attack success rate indicates low model security, necessitating developers to employ adversarial training or other methods to enhance model security. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

TransFlowGAN: Generation and Balancing of Network Traffic Data 24

TransFlowGAN: Generation and Balancing of Network Traffic Da...

引用

2024 Asia Pacific Conference on Computing Technologies, Communications and Networking, CTCNet 2024

作者： Zhang, Hongzheng National Key Laboratory of Science and Technology on Information System Security China

ISBN: (纸本)9798400709609

In recent years, with the rapid development of deep learning technology, researchers in the field of network security have begun to explore the use of deep learning to solve the problem of encrypted traffic classification. However, existing publicly available encrypted traffic datasets suffer from severe class imbalance issues, which significantly impact the performance of deep learning classification methods. Constructing a complete encrypted traffic dataset from scratch is both time-consuming and costly. To address this issue, this paper proposes an innovative model named TransFlowRGAN, which is an encrypted traffic sequence feature generation model that combines Transformer and Relative Generative Adversarial Network (RGAN). Our model leverages the advantages of RGAN, effectively mitigating the instability and mode collapse issues that traditional GAN models often encounter during training. Additionally, it incorporates the Transformer architecture. The self-attention mechanism of the Transformer can better capture the complex features and long-term dependencies in traffic data, resulting in the generation of higher-quality data. We evaluated the performance of our model on the ISCXVPN2016 public dataset. The results demonstrate that TransFlowRGAN excels in handling imbalanced datasets, significantly outperforming other data augmentation models. © 2024 Copyright held by the owner/author(s).

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Feature Distillation GAN: An Adversarial Example Defense Method Based on Feature Reconstruction 2nd

Feature Distillation GAN: An Adversarial Example Defense Met...

引用

2nd International Conference on Network Simulation and Evaluation, NSE 2023

作者： Liu, Ruiqi Li, Hu Wang, Dongxia Zhao, Chongyang Li, Boyu National Key Laboratory of Science and Technology On Information System Security Beijing China

ISBN: (纸本)9789819745180

Purifying and reconstructing adversarial examples into benign ones by generative models is a class of defense methods to eliminate adversarial perturbations. These methods do not make assumptions on adversarial attacks or model architectures, and thus can defend classifiers against unseen threats. In this work, we propose a new GAN-based adversarial purification method Feature Distillation GAN (FDGAN). We consider how to eliminate the feature differences between adversarial examples and benign examples which are difficult for humans to perceive. We extract and analyze the differences between the adversarial examples and the benign examples from multiple dimensionals. On the basis of comparing the differences among various features, we propose a new feature-level perspective on adversarial purification, design a new adversarial example purification method for image inputs, and implement our method based on GAN. Experimental results show that FDGAN is effective against unknown adversarial attacks and outperforms all the purification methods we used for comparison. Evaluating FDGAN against strong adaptive attacks, the robust accuracy of FDGAN under AutoAttack (L∞ = 8/255) exceeds the SOTA adversarial training methods on the Robustbench. Furthermore, FDGAN has lower training cost and can be combined with a variety of model-based defense strategies to further improve the defense effectiveness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

IMCSCL: Image-Based Malware Classification using Self-Supervised and Contrastive Learning 23

IMCSCL: Image-Based Malware Classification using Self-Superv...

引用

23rd IEEE International Conference on Software Quality, Reliability, and security, QRS 2023

作者： Jia, Yetao Meng, Yangyang Zhuang, Honglin National Key Laboratory of Science and Technology on Information System Security Beijing China

ISBN: (纸本)9798350319583

The use of malware for illicit cyber activities, including network attacks and information theft, poses a severe threat to cybersecurity. In comparison to traditional malware detection methods based on signature and heuristics, machine learning and deep learning-based malware detection methods demonstrate superior generalization ability. However, existing research still faces challenges such as reliance on relatively single malware features, inadequate ability to describe malware features, and overdependence on labeled data. In this paper, we propose an image-based malware classification method using self-supervised and contrastive learning, named IMCSCL. We visualize malware using opcode semantic features, and then detect malware using a contrastive learning method with improved feature encoder network. Experimental results demonstrate that IMCSCL achieves higher detection accuracy compared to supervised malware detection methods, achieving 98.85% accuracy on the Microsoft Malware Classification Challenge dataset. Fine-tuning the model using randomly selected 5% labeled samples from the training set still achieved high accuracy of 94.22%. IMCSCL exhibits superior generalization ability, faster convergence speed, and better training stability. Moreover, contrastive learning significantly reduces malware labeling costs while effectively enhancing detection performance. © 2023 IEEE.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Adversarial Example Detection for Deep Neural Networks: A Review 8

Adversarial Example Detection for Deep Neural Networks: A Re...

引用

8th International Conference on Data science in Cyberspace, DSC 2023

作者： Zhao, Chongyang Li, Hu Wang, Dongxia Liu, Ruiqi National Key Laboratory of Science and Technology on Information System Security Beijing China

ISBN: (纸本)9798350331035

Deep neural networks have been widely applied in various critical domains. However, they are vulnerable to the threat of adversarial examples. It is challenging to make deep neural networks inherently robust to adversarial examples, while adversarial example detection offers advantages such as not affecting model classification accuracy. This paper introduces common adversarial attack methods and provides an explanation of adversarial example detection. Recent advances in adversarial example detection methods are categorized into two major classes: statistical methods and adversarial detection networks. The evolutionary relationship among different detection methods is discussed. Finally, the current research status in this field is summarized, and potential future directions are highlighted. © 2023 IEEE.

关键词： adversarial detection adversarial example deep neural network

来源：评论

学校读者我要写书评

暂无评论

m2Vib: 2D Vibration Measurement Using a Single mmWave Radar 30

m2Vib: 2D Vibration Measurement Using a Single mmWave Radar

引用

30th IEEE International Conference on Parallel and Distributed systems, ICPADS 2024

作者： Wang, Haolin Xu, Wei Li, Zezhao Yang, Yanni Huang, Zhijian Shandong University School of Computer Science and Technology Qingdao China National Key Laboratory of Science and Technology on Information System Security China

ISBN: (纸本)9798331515966

Two-dimensional vibration measurement is essential for monitoring structure and machine health, especially providing more information than those 1D solutions. However, most existing 2D vibration monitoring methods require multi-dimension sensors, which can be costly for large-scale deployment. In this paper, we propose m^2 Vib, the first work that can realize accurate 2D vibration measurement using a single mmWave radar. Our system is achieved by creating a virtual radar with non-line-of-sight (NLoS) signals from a nearby reflector (e.g., wall) to offer the 2nd-dimension vibration observations. We further establish a geometric model to unveil the relationship between 2D vibration displacement and mmWave signal. Such an idea still encounters challenges from significantly attenuated NLoS signals and low-resolution geometric parameters from the commercial mmWave radar. Thus, we design a new retroreflector to enhance NLoS signals and advanced signal processing methods to obtain fine-grained geometric parameters. Extensive experimental results show that we can achieve sub-mm level 2D vibration measurement with an average vibration displacement error of 62.2 um and frequency error of 0.01 Hz in the X-axis and Y-axis. © 2024 IEEE.

关键词： Vibration measurement

来源：评论

学校读者我要写书评

暂无评论

Efficient Penetration Testing Path Planning Based on Reinforcement Learning with Episodic Memory

引用

Computer Modeling in Engineering & sciences 2024年第9期140卷 2613-2634页

作者： Ziqiao Zhou Tianyang Zhou Jinghao Xu Junhu Zhu Henan Key Laboratory of Information Security National Engineering Technology Research Center of the Digital Switching SystemZhengzhou450000China School of Cryptographic Engineering Information Engineering UniversityZhengzhou450000China

Intelligent penetration testing is of great significance for the improvement of the security of information systems,and the critical issue is the planning of penetration test *** view of the difficulty for attackers to obtain complete network information in realistic network scenarios,Reinforcement Learning(RL)is a promising solution to discover the optimal penetration path under incomplete information about the target *** RL-based methods are challenged by the sizeable discrete action space,which leads to difficulties in the ***,most methods still rely on experts’*** address these issues,this paper proposes a penetration path planning method based on reinforcement learning with episodic ***,the penetration testing problem is formally described in terms of reinforcement *** speed up the training process without specific prior knowledge,the proposed algorithm introduces episodic memory to store experienced advantageous strategies for the first ***,the method offers an exploration strategy based on episodic memory to guide the agents in *** design makes full use of historical experience to achieve the purpose of reducing blind exploration and improving planning ***,comparison experiments are carried out with the existing RL-based *** results reveal that the proposed method has better convergence *** running time is reduced by more than 20%.

关键词： Intelligent penetration testing penetration testing path planning reinforcement learning episodic memory exploration strategy

来源：评论

学校读者我要写书评

暂无评论

Fast measurement and prediction method for electromagnetic susceptibility of receiver

引用

Journal of systems Engineering and Electronics 2024年第2期35卷 275-285页

作者： CHEN Yan LU Zhonghao LIU Yunxia State Key Laboratory of Complex Electromagnetic Environment Effects on Electronics and Information System College of Electronic Science and TechnologyNational University of Defense TechnologyChangsha 410073China State Key Laboratory of Astronautic Dynamics Xi’an 710043China

Aiming at evaluating and predicting rapidly and accurately a high sensitivity receiver’s adaptability in complex electromagnetic environments,a novel testing and prediction method based on dual-channel multi-frequency is proposed to improve the traditional two-tone ***,two signal generators are used to generate signals at the radio frequency(RF)by frequency scanning,and then a rapid measurement at the intermediate frequency(IF)output port is carried out to obtain a huge amount of sample data for the subsequent ***,the IF output response data are modeled and analyzed to construct the linear and nonlinear response constraint equations in the frequency domain and prediction models in the power domain,which provide the theoretical criteria for interpreting and predicting electromagnetic susceptibility(EMS)of the *** experiment performed on a radar receiver confirms the reliability of the method proposed in this *** shows that the interference of each harmonic frequency and each order to the receiver can be identified and predicted with the sensitivity *** on this,fast and comprehensive evaluation and prediction of the receiver’s EMS in complex environment can be efficiently realized.

关键词： electromagnetic susceptibility(EMS) receiver dualchannel multi-frequency nonlinear response frequency domain power domain

来源：评论

学校读者我要写书评

暂无评论

In-memory computing based on phase change memory for high energy efficiency

引用

science China(information sciences) 2023年第10期66卷 20-41页

作者： Luchang HE Xi LI Chenchen XIE Zhitang SONG National Key Laboratory of Materials for Integrated Circuits Shanghai Institute of Micro-System and Information Technology Chinese Academy of Sciences School of Microelectronics University of Science and Technology of China

The energy efficiency issue caused by the memory wall in traditional von Neumann architecture is difficult to reconcile. In-memory computing(CIM) based on emerging nonvolatile memory(NVM) is a promising solution to avoid data movement between storage and processors and realize highly energy-efficient computing. Compared with other NVM technologies, phase change random access memory(PCM) exhibits comprehensive performance for analog computing. In this paper, we review advanced PCM techniques,including phase-change materials, mechanisms, and unique properties, as a foundation and inspiration for implementing CIM architecture. Meanwhile, state-of-the-art PCM-based CIM systems are well discussed for high energy efficiency in artificial neural networks, spiking neural networks, and other artificial intelligence(AI) applications. Finally, we present the remaining challenges and potential solutions of CIM for further investigation.

关键词： in-memory computing phase change memory energy efficiency neural network memory wall

来源：评论

学校读者我要写书评

暂无评论

Machine Learning security Analysis Framework Based on Threat Matrix 8

Machine Learning Security Analysis Framework Based on Threat...

引用

8th International Conference on Data science in Cyberspace, DSC 2023

作者： Li, Hu Ma, Qiongmin Zhang, Ming Kuang, Xiaohui National Key Laboratory of Science and Technology on Information System Security Beijing China National Key Laboratory for Complex Systems Simulation Beijing China

ISBN: (纸本)9798350331035

Machine Learning and Artificial Intelligence technology accelerates technological progress and promotes social development, but also brings many security problems. Machine learning models may be affected, deceived, controlled or destroyed at different stages, e.g. training and inference, which may lead to serious consequences. In view of the main stages of the life cycle of machine learning, we summarizes the related attack and defense technologies based on threat matrix, makes a comprehensive and systematic analysis on the security issues of machine learning, and puts forward several security protection suggestions for intelligent information systems. © 2023 IEEE.

关键词： Artificial Intelligence Machine Learning security Analysis Threat Matrix

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：