Adversarial attacks against conventional Deep Learning (DL) systems and algorithms have been widely studied, and various defenses were proposed. However, the possibility and feasibility of such attacks against Deep Reinforcement Learning (DRL) are less explored. As DRL has achieved great success in various complex tasks, designing effective adversarial attacks is an indispensable prerequisite towards building robust DRL algorithms. In this paper, we introduce two novel adversarial attack techniques to \emph{stealthily} and \emph{efficiently} attack the DRL agents. These two techniques enable an adversary to inject adversarial samples in a minimal set of critical moments while causing the most severe damage to the agent. The first technique is the \emph{critical point attack}: the adversary builds a model to predict the future environmental states and agent's actions, assesses the damage of each possible attack strategy, and selects the optimal one. The second technique is the \emph{antagonist attack}: the adversary automatically learns a domain-agnostic model to discover the critical moments of attacking the agent in an episode. Experimental results demonstrate the effectiveness of our techniques. Specifically, to successfully attack the DRL agent, our critical point technique only requires 1 (TORCS) or 2 (Atari Pong and Breakout) steps, and the antagonist technique needs fewer than 5 steps (4 Mujoco tasks), which are significant improvements over state-of-the-art methods.
Technische Universität Chemnitz（开姆尼茨工业大学）；Hanoi University of Science and Technology（河内理工大学）
Software Defined Networking (SDN) enables flexible and scalable network control and management. However, it also introduces new vulnerabilities that can be exploited by attackers. In particular, low-rate and slow or stealthy Denial-of-Service (DoS) attacks are recently attracting attention from researchers because of their detection challenges. In this paper, we propose a novel machine learning based defense framework named Q-MIND, to effectively detect and mitigate stealthy DoS attacks in SDN-based networks. We first analyze the adversary model of stealthy DoS attacks, the related vulnerabili…
Graduate School of Computer Science, Saarland University（萨尔大学计算机科学研究生院）；Bosch Center for Artificial Intelligence, Robert Bosch GmbH（罗伯特·博世有限公司博世人工智能中心）；CISPA Helmholtz Center for Information Security（CISPA亥姆霍兹信息安全中心）
Stealthy Poisoning Attacks Bypass Defenses in Regression Settings
IBM Research Europe, Portal, First Floor Trinity Business School, Trinity College Dublin（IBM欧洲研究院，波特尔大楼，特里尼蒂商务学校，特里尼蒂学院都柏林）；Imperial College London, South Kensington Campus（伦敦帝国理工学院，南肯辛顿校区）；Universidad de Alcalá de Henares, Escuela Politécnica Superior（阿尔卡萨大学，工程技术学院）；Defence Science and Technology Laboratory (DSTL), Porton Down, Salisbury（国防科学与技术实验室（DSTL），波特恩道，萨里）
To reduce the time-to-market and access to state-of-the-art techniques, CNN hardware mapping and deployment on embedded accelerators are often outsourced to untrusted third parties, which is going to be more prevalent in futuristic artificial intelligence of things (AIoT) systems. These AIoT systems anticipate horizontal collaboration among different resource-constrained AIoT node devices, where CNN layers are partitioned and these devices collaboratively compute complex CNN tasks. This horizontal collaboration opens another attack surface to the CNN-based application, like inserting the hardw…
Zhejiang University of Technology（浙江工业大学）；Zhejiang University（浙江大学）
Deep neural networks are susceptible to poisoning attacks by purposely polluted training data with specific triggers. As existing episodes mainly focused on attack success rate with patch-based samples, defense algorithms can easily detect these poisoning samples. We propose DeepPoison, a novel adversarial network of one generator and two discriminators, to address this problem. Specifically, the generator automatically extracts the target class' hidden features and embeds them into benign training samples. One discriminator controls the ratio of the poisoning perturbation. The other discrimin…
Deep neural networks are vulnerable to Trojan attacks. Existing attacks use visible patterns (e.g., a patch or image transformations) as triggers, which are vulnerable to human inspection. In this paper, we propose stealthy and efficient Trojan attacks, BppAttack. Based on existing biology literature on human visual systems, we propose to use image quantization and dithering as the Trojan trigger, making imperceptible changes. It is a stealthy and efficient attack without training auxiliary models. Due to the small changes made to images, it is hard to inject such triggers during training. To …
It has been shown that natural language processing (NLP) models are vulnerable to a kind of security threat called the Backdoor Attack, which utilizes a `backdoor trigger' paradigm to mislead the models. The most threatening backdoor attack is the stealthy backdoor, which defines the triggers as text style or syntactic. Although they have achieved an incredible high attack success rate (ASR), we find that the principal factor contributing to their ASR is not the `backdoor trigger' paradigm. Thus the capacity of these stealthy backdoor attacks is overestimated when categorized as backdoor attac…
Stealthy attacks on Industrial Control Systems can cause significant damage while evading detection. In this paper, instead of focusing on the detection of stealthy attacks, we aim to provide early warnings to operators, in order to avoid physical damage and preserve in advance data that may serve as an evidence during an investigation. We propose a framework to provide grounds for suspicion, i.e. preliminary indicators reflecting the likelihood of success of a stealthy attack. We propose two grounds for suspicion based on the behaviour of the physical process: (i) feasibility of a stealthy at…
Shadowcast: Stealthy Data Poisoning Attacks Against Vision-Language Models
University of Maryland, College Park（马里兰大学帕克分校）；University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Salesforce Research（Salesforce研究院）；Apple（苹果公司）；Netflix（网飞公司）；Eyeline Studios（Eyeline工作室）；University of Waterloo（滑铁卢大学）
For a safe and reliable operation of the smart grid, timely detection of cyber-attacks is of critical importance. Moreover, considering smarter and more capable attackers, robust detection mechanisms are needed against a diverse range of cyber-attacks. With these purposes, we propose a robust online detection algorithm for (possibly combined) false data injection (FDI) and jamming attacks, that also provides online estimates of the unknown and time-varying attack parameters and recovered state estimates. Further, considering smarter attackers that are capable of designing stealthy attacks to p…
Recently, deep neural networks (DNNs) have been deployed in safety-critical systems such as autonomous vehicles and medical devices. Shortly after that, the vulnerability of DNNs were revealed by stealthy adversarial examples where crafted inputs -- by adding tiny perturbations to original inputs -- can lead a DNN to generate misclassification outputs. To improve the robustness of DNNs, some algorithmic-based countermeasures against adversarial examples have been introduced thereafter. In this paper, we propose a new type of stealthy attack on protected DNNs to circumvent the algorithmic defen…
Federated learning (FL) has become an emerging machine learning technique lately due to its efficacy in safeguarding the client's confidential information. Nevertheless, despite the inherent and additional privacy-preserving mechanisms (e.g., differential privacy, secure multi-party computation, etc.), the FL models are still vulnerable to various privacy-violating and security-compromising attacks (e.g., data or model poisoning) due to their numerous attack vectors which in turn, make the models either ineffective or sub-optimal. Existing adversarial models focusing on untargeted model poison…
Industrial control systems (ICS), which in many cases are components of critical national infrastructure, are increasingly being connected to other networks and the wider internet motivated by factors such as enhanced operational functionality and improved efficiency. However, set in this context, it is easy to see that the cyber attack surface of these systems is expanding, making it more important than ever that innovative solutions for securing ICS be developed and that the limitations of these solutions are well understood. The development of anomaly based intrusion detection techniques ha…
Towards Stealthy Backdoor Attacks against Speech Recognition via Elements of Sound
Hohai University（河海大学）；RMIT University（皇家墨尔本理工大学）；Sun Yat-sen University（中山大学）；Delft University of Technology（代尔夫特理工大学）；Tsinghua University（清华大学）；Ant Group（蚂蚁集团）
STAIN-FL: Stealthy Targeted Attack Injection with Contextual Triggers in Federated Learning
Singapore Institute of Technology (SIT)（新加坡理工学院（SIT））；SingHealth Duke-NUS AI in Medicine Institute（新加坡保健服务集团杜克-新加坡国立大学医学院AI研究所）；SingHealth AI Office（新加坡保健服务集团AI办公室）
Internet of Things (IoT) networks consist of sensors, actuators, mobile and wearable devices that can connect to the Internet. With billions of such devices already in the market which have significant vulnerabilities, there is a dangerous threat to the Internet services and also some cyber-physical systems that are also connected to the Internet. Specifically, due to their existing vulnerabilities IoT devices are susceptible to being compromised and being part of a new type of stealthy Distributed Denial of Service (DDoS) attack, called Mongolian DDoS, which is characterized by its widely dis…
Shanghai Jiao Tong University（上海交通大学）；Beijing University of Posts and Telecommunications（北京邮电大学）；Tsinghua University（清华大学）；Zhongguancun Laboratory（中关村实验室）；Nanyang Technological University（南洋理工大学）；Alibaba Group（阿里巴巴集团）
Transforming off-the-shelf deep neural network (DNN) models into dynamic multi-exit architectures can achieve inference and transmission efficiency by fragmenting and distributing a large DNN model in edge computing scenarios (e.g., edge devices and cloud servers). In this paper, we propose a novel backdoor attack specifically on the dynamic multi-exit DNN models. Particularly, we inject a backdoor by poisoning one DNN model's shallow hidden layers targeting not this vanilla DNN model but only its dynamically deployed multi-exit architectures. Our backdoored vanilla model behaves normally on p…
The false data injection (FDI) attack cannot be detected by the traditional anomaly detection techniques used in the energy system state estimators. In this paper, we demonstrate how FDI attacks can be constructed blindly, i.e., without system knowledge, including topological connectivity and line reactance information. Our analysis reveals that existing FDI attacks become detectable (consequently unsuccessful) by the state estimator if the data contains grossly corrupted measurements such as device malfunction and communication errors. The proposed sparse optimization based stealthy attacks c…
Due to the importance of Critical Infrastructure (CI) in a nation's economy, they have been lucrative targets for cyber attackers. These critical infrastructures are usually Cyber-Physical Systems (CPS) such as power grids, water, and sewage treatment facilities, oil and gas pipelines, etc. In recent times, these systems have suffered from cyber attacks numerous times. Researchers have been developing cyber security solutions for CIs to avoid lasting damages. According to standard frameworks, cyber security based on identification, protection, detection, response, and recovery are at the core …
Attacks on Industrial Control Systems (ICS) can lead to significant physical damage. While offline safety and security assessments can provide insight into vulnerable system components, they may not account for stealthy attacks designed to evade anomaly detectors during long operational transients. In this paper, we propose a predictive online monitoring approach to check the safety of the system under potential stealthy attacks. Specifically, we adapt previous results in reachability analysis for attack impact assessment to provide an efficient algorithm for online safety monitoring for Linea…
Nowadays, vehicles have complex in-vehicle networks (IVNs) with millions of lines of code controlling almost every function in the vehicle including safety-critical functions. It has recently been shown that IVNs are becoming increasingly vulnerable to cyber-attacks capable of taking control of vehicles, thereby threatening the safety of the passengers. Several countermeasures have been proposed in the literature in response to the arising threats, however, hurdle requirements imposed by the industry is hindering their adoption in practice. In particular, detecting attacks on IVNs is challenge…
Defenses against security threats have been an interest of recent studies. Recent works have shown that it is not difficult to attack a natural language processing (NLP) model while defending against them is still a cat-mouse game. Backdoor attacks are one such attack where a neural network is made to perform in a certain way on specific samples containing some triggers while achieving normal results on other samples. In this work, we present a few defense strategies that can be useful to counter against such an attack. We show that our defense methodologies significantly decrease the performa…
Research on cache attacks has shown that CPU caches leak significant information. Proposed detection mechanisms assume that all cache attacks cause more cache hits and cache misses than benign applications and use hardware performance counters for detection. In this article, we show that this assumption does not hold by developing a novel attack technique: the Flush+Flush attack. The Flush+Flush attack only relies on the execution time of the flush instruction, which depends on whether data is cached or not. Flush+Flush does not make any memory accesses, contrary to any other cache attack. Thu…
In this paper, we propose an effective and easily deployable approach to detect the presence of stealthy sensor attacks in industrial control systems, where (legacy) control devices critically rely on accurate (and usually non-encrypted) sensor readings. Specifically, we focus on stealthy attacks that crash a sensor and then immediately impersonate that sensor by sending out fake readings. We consider attackers who aim to stay hidden in the system for a prolonged period. To detect such attacks, our approach relies on continuous injection of "micro distortion" to the original sensor's readings.…
Harmless Yet Harmful: Neutral Prompting Attacks for Stealthy Hallucination Steering in Agent Skills
Department of Computer Science（计算机科学系）；National Yang Ming Chiao Tung University（阳明交通大学）；Department of Electronics and Electrical Engineering（电子与电气工程系）；School of Computing（计算学院）；Institute of Science Tokyo（东京科学研究所）
Stealthy Yet Effective: Distribution-Preserving Backdoor Attacks on Graph Classification
College of Intelligence and Computing, Tianjin University, Tianjin, China（天津大学智能与计算学院）；Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ), Shenzhen, China（广东人工智能与数字经济实验室（深圳））；College of Artificial Intelligence, Dalian Maritime University, Dalian, China（大连海事大学人工智能学院）
Hammering the Diagnosis: Rowhammer-Induced Stealthy Trojan Attacks on ViT-Based Medical Imaging
Department of Electrical and Computer Engineering, University of Arizona, Tucson, AZ, USA（亚利桑那大学电气与计算机工程系）；Department of Electrical and Computer Engineering, University of California, Davis, CA, USA（加州大学戴维斯分校电气与计算机工程系）；Department of Computer Engineering and Computer Science, California State University, Long Beach, Long Beach, CA, USA（加州州立大学-long beach分校计算机工程与科学系）
Deep learning architectures (DLA) have shown impressive performance in computer vision, natural language processing and so on. Many DLA make use of cloud computing to achieve classification due to the high computation and memory requirements. Privacy and latency concerns resulting from cloud computing has inspired the deployment of DLA on embedded hardware accelerators. To achieve short time-to-market and have access to global experts, state-of-the-art techniques of DLA deployment on hardware accelerators are outsourced to untrusted third parties. This outsourcing raises security concerns as h…
