Adversarial distillation(AD)has emerged as a potential solution to tackle the challenging optimization problem of loss with hard labels in adversarial training.However,fixed sample-agnostic and student-egocentric atta...Adversarial distillation(AD)has emerged as a potential solution to tackle the challenging optimization problem of loss with hard labels in adversarial training.However,fixed sample-agnostic and student-egocentric attack strategies are unsuitable for distillation.Additionally,the reliability of guidance from static teachers diminishes as target models become more robust.This paper proposes an AD method called Learnable Distillation Attack Strategies and Evolvable Teachers Adversarial Distillation(LDAS&ET-AD).Firstly,a learnable distillation attack strategies generating mechanism is developed to automatically generate sample-dependent attack strategies tailored for distillation.A strategy model is introduced to produce attack strategies that enable adversarial examples(AEs)to be created in areas where the target model significantly diverges from the teachers by competing with the target model in minimizing or maximizing the AD loss.Secondly,a teacher evolution strategy is introduced to enhance the reliability and effectiveness of knowledge in improving the generalization performance of the target model.By calculating the experimentally updated target model’s validation performance on both clean samples and AEs,the impact of distillation from each training sample and AE on the target model’s generalization and robustness abilities is assessed to serve as feedback to fine-tune standard and robust teachers accordingly.Experiments evaluate the performance of LDAS&ET-AD against different adversarial attacks on the CIFAR-10 and CIFAR-100 datasets.The experimental results demonstrate that the proposed method achieves a robust precision of 45.39%and 42.63%against AutoAttack(AA)on the CIFAR-10 dataset for ResNet-18 and MobileNet-V2,respectively,marking an improvement of 2.31%and 3.49%over the baseline method.In comparison to state-of-the-art adversarial defense techniques,our method surpasses Introspective Adversarial Distillation,the top-performing method in terms of robustness under AA attack for the CIFAR-10 dataset,with enhancements of 1.40%and 1.43%for ResNet-18 and MobileNet-V2,respectively.These findings demonstrate the effectiveness of our proposed method in enhancing the robustness of deep learning networks(DNNs)against prevalent adversarial attacks when compared to other competing methods.In conclusion,LDAS&ET-AD provides reliable and informative soft labels to one of the most promising defense methods,AT,alleviating the limitations of untrusted teachers and unsuitable AEs in existing AD techniques.We hope this paper promotes the development of DNNs in real-world trust-sensitive fields and helps ensure a more secure and dependable future for artificial intelligence systems.展开更多
A kind of attack strategy based on a probabilistic cloning machine is proposed in this letter. The security of BB84 and the six-state quantum key distribution protocols under this attack is studied by theoretic analys...A kind of attack strategy based on a probabilistic cloning machine is proposed in this letter. The security of BB84 and the six-state quantum key distribution protocols under this attack is studied by theoretic analyses and corroborated by simulations. It is concluded that the quantum key distribution protocols still have an asymptotic perfect security even if the eavesdropper adopts the proposed attack strategy.展开更多
Cloud computing involves remote server deployments with public net-work infrastructures that allow clients to access computational resources.Virtual Machines(VMs)are supplied on requests and launched without interacti...Cloud computing involves remote server deployments with public net-work infrastructures that allow clients to access computational resources.Virtual Machines(VMs)are supplied on requests and launched without interactions from service providers.Intruders can target these servers and establish malicious con-nections on VMs for carrying out attacks on other clustered VMs.The existing system has issues with execution time and false-positive rates.Hence,the overall system performance is degraded considerably.The proposed approach is designed to eliminate Cross-VM side attacks and VM escape and hide the server’s position so that the opponent cannot track the target server beyond a certain point.Every request is passed from source to destination via one broadcast domain to confuse the opponent and avoid them from tracking the server’s position.Allocation of SECURITY Resources accepts a safety game in a simple format as input andfinds the best coverage vector for the opponent using a Stackelberg Equilibrium(SSE)technique.A Mixed Integer Linear Programming(MILP)framework is used in the algorithm.The VM challenge is reduced by afirewall-based controlling mechanism combining behavior-based detection and signature-based virus detection.The pro-posed method is focused on detecting malware attacks effectively and providing better security for the VMs.Finally,the experimental results indicate that the pro-posed security method is efficient.It consumes minimum execution time,better false positive rate,accuracy,and memory usage than the conventional approach.展开更多
纵向联邦学习(Vertical federated learning,VFL)是一种新兴的分布式机器学习技术,在保障隐私性的前提下,利用分散在各个机构的数据实现机器学习模型的联合训练.纵向联邦学习被广泛应用于工业互联网、金融借贷和医疗诊断等诸多领域中,...纵向联邦学习(Vertical federated learning,VFL)是一种新兴的分布式机器学习技术,在保障隐私性的前提下,利用分散在各个机构的数据实现机器学习模型的联合训练.纵向联邦学习被广泛应用于工业互联网、金融借贷和医疗诊断等诸多领域中,因此保证其隐私安全性具有重要意义.首先,针对纵向联邦学习协议中由于参与方交换的嵌入表示造成的隐私泄漏风险,研究由协作者发起的通用的属性推断攻击.攻击者利用辅助数据和嵌入表示训练一个攻击模型,然后利用训练完成的攻击模型窃取参与方的隐私属性.实验结果表明,纵向联邦学习在训练推理阶段产生的嵌入表示容易泄漏数据隐私.为了应对上述隐私泄漏风险,提出一种基于最大−最小策略的纵向联邦学习隐私保护方法(Privacy preservation method for vertical federated learning based on max-min strategy,PPVFL),其引入梯度正则组件保证训练过程主任务的预测性能,同时引入重构组件掩藏参与方嵌入表示中包含的隐私属性信息.最后,在钢板缺陷诊断工业场景的实验结果表明,相比于没有任何防御方法的VFL,隐私保护方法将攻击推断准确度从95%下降到55%以下,接近于随机猜测的水平,同时主任务预测准确率仅下降2%.展开更多
The rapid integration of artificial intelligence (AI) into critical sectors has revealed a complex landscape of cybersecurity challenges that are unique to these advanced technologies. AI systems, with their extensive...The rapid integration of artificial intelligence (AI) into critical sectors has revealed a complex landscape of cybersecurity challenges that are unique to these advanced technologies. AI systems, with their extensive data dependencies and algorithmic complexities, are susceptible to a broad spectrum of cyber threats that can undermine their functionality and compromise their integrity. This paper provides a detailed analysis of these threats, which include data poisoning, adversarial attacks, and systemic vulnerabilities that arise from the AI’s operational and infrastructural frameworks. This paper critically examines the effectiveness of existing defensive mechanisms, such as adversarial training and threat modeling, that aim to fortify AI systems against such vulnerabilities. In response to the limitations of current approaches, this paper explores a comprehensive framework for the design and implementation of robust AI systems. This framework emphasizes the development of dynamic, adaptive security measures that can evolve in response to new and emerging cyber threats, thereby enhancing the resilience of AI systems. Furthermore, the paper addresses the ethical dimensions of AI cybersecurity, highlighting the need for strategies that not only protect systems but also preserve user privacy and ensure fairness across all operations. In addition to current strategies and ethical concerns, this paper explores future directions in AI cybersecurity.展开更多
基金the National Key Research and Development Program of China(2021YFB1006200)Major Science and Technology Project of Henan Province in China(221100211200).Grant was received by S.Li.
文摘Adversarial distillation(AD)has emerged as a potential solution to tackle the challenging optimization problem of loss with hard labels in adversarial training.However,fixed sample-agnostic and student-egocentric attack strategies are unsuitable for distillation.Additionally,the reliability of guidance from static teachers diminishes as target models become more robust.This paper proposes an AD method called Learnable Distillation Attack Strategies and Evolvable Teachers Adversarial Distillation(LDAS&ET-AD).Firstly,a learnable distillation attack strategies generating mechanism is developed to automatically generate sample-dependent attack strategies tailored for distillation.A strategy model is introduced to produce attack strategies that enable adversarial examples(AEs)to be created in areas where the target model significantly diverges from the teachers by competing with the target model in minimizing or maximizing the AD loss.Secondly,a teacher evolution strategy is introduced to enhance the reliability and effectiveness of knowledge in improving the generalization performance of the target model.By calculating the experimentally updated target model’s validation performance on both clean samples and AEs,the impact of distillation from each training sample and AE on the target model’s generalization and robustness abilities is assessed to serve as feedback to fine-tune standard and robust teachers accordingly.Experiments evaluate the performance of LDAS&ET-AD against different adversarial attacks on the CIFAR-10 and CIFAR-100 datasets.The experimental results demonstrate that the proposed method achieves a robust precision of 45.39%and 42.63%against AutoAttack(AA)on the CIFAR-10 dataset for ResNet-18 and MobileNet-V2,respectively,marking an improvement of 2.31%and 3.49%over the baseline method.In comparison to state-of-the-art adversarial defense techniques,our method surpasses Introspective Adversarial Distillation,the top-performing method in terms of robustness under AA attack for the CIFAR-10 dataset,with enhancements of 1.40%and 1.43%for ResNet-18 and MobileNet-V2,respectively.These findings demonstrate the effectiveness of our proposed method in enhancing the robustness of deep learning networks(DNNs)against prevalent adversarial attacks when compared to other competing methods.In conclusion,LDAS&ET-AD provides reliable and informative soft labels to one of the most promising defense methods,AT,alleviating the limitations of untrusted teachers and unsuitable AEs in existing AD techniques.We hope this paper promotes the development of DNNs in real-world trust-sensitive fields and helps ensure a more secure and dependable future for artificial intelligence systems.
文摘A kind of attack strategy based on a probabilistic cloning machine is proposed in this letter. The security of BB84 and the six-state quantum key distribution protocols under this attack is studied by theoretic analyses and corroborated by simulations. It is concluded that the quantum key distribution protocols still have an asymptotic perfect security even if the eavesdropper adopts the proposed attack strategy.
文摘Cloud computing involves remote server deployments with public net-work infrastructures that allow clients to access computational resources.Virtual Machines(VMs)are supplied on requests and launched without interactions from service providers.Intruders can target these servers and establish malicious con-nections on VMs for carrying out attacks on other clustered VMs.The existing system has issues with execution time and false-positive rates.Hence,the overall system performance is degraded considerably.The proposed approach is designed to eliminate Cross-VM side attacks and VM escape and hide the server’s position so that the opponent cannot track the target server beyond a certain point.Every request is passed from source to destination via one broadcast domain to confuse the opponent and avoid them from tracking the server’s position.Allocation of SECURITY Resources accepts a safety game in a simple format as input andfinds the best coverage vector for the opponent using a Stackelberg Equilibrium(SSE)technique.A Mixed Integer Linear Programming(MILP)framework is used in the algorithm.The VM challenge is reduced by afirewall-based controlling mechanism combining behavior-based detection and signature-based virus detection.The pro-posed method is focused on detecting malware attacks effectively and providing better security for the VMs.Finally,the experimental results indicate that the pro-posed security method is efficient.It consumes minimum execution time,better false positive rate,accuracy,and memory usage than the conventional approach.
文摘纵向联邦学习(Vertical federated learning,VFL)是一种新兴的分布式机器学习技术,在保障隐私性的前提下,利用分散在各个机构的数据实现机器学习模型的联合训练.纵向联邦学习被广泛应用于工业互联网、金融借贷和医疗诊断等诸多领域中,因此保证其隐私安全性具有重要意义.首先,针对纵向联邦学习协议中由于参与方交换的嵌入表示造成的隐私泄漏风险,研究由协作者发起的通用的属性推断攻击.攻击者利用辅助数据和嵌入表示训练一个攻击模型,然后利用训练完成的攻击模型窃取参与方的隐私属性.实验结果表明,纵向联邦学习在训练推理阶段产生的嵌入表示容易泄漏数据隐私.为了应对上述隐私泄漏风险,提出一种基于最大−最小策略的纵向联邦学习隐私保护方法(Privacy preservation method for vertical federated learning based on max-min strategy,PPVFL),其引入梯度正则组件保证训练过程主任务的预测性能,同时引入重构组件掩藏参与方嵌入表示中包含的隐私属性信息.最后,在钢板缺陷诊断工业场景的实验结果表明,相比于没有任何防御方法的VFL,隐私保护方法将攻击推断准确度从95%下降到55%以下,接近于随机猜测的水平,同时主任务预测准确率仅下降2%.
文摘The rapid integration of artificial intelligence (AI) into critical sectors has revealed a complex landscape of cybersecurity challenges that are unique to these advanced technologies. AI systems, with their extensive data dependencies and algorithmic complexities, are susceptible to a broad spectrum of cyber threats that can undermine their functionality and compromise their integrity. This paper provides a detailed analysis of these threats, which include data poisoning, adversarial attacks, and systemic vulnerabilities that arise from the AI’s operational and infrastructural frameworks. This paper critically examines the effectiveness of existing defensive mechanisms, such as adversarial training and threat modeling, that aim to fortify AI systems against such vulnerabilities. In response to the limitations of current approaches, this paper explores a comprehensive framework for the design and implementation of robust AI systems. This framework emphasizes the development of dynamic, adaptive security measures that can evolve in response to new and emerging cyber threats, thereby enhancing the resilience of AI systems. Furthermore, the paper addresses the ethical dimensions of AI cybersecurity, highlighting the need for strategies that not only protect systems but also preserve user privacy and ensure fairness across all operations. In addition to current strategies and ethical concerns, this paper explores future directions in AI cybersecurity.
