This paper proposes a new defense against neural network backdooring attacks that are maliciously trained to mispredict in the presence of attacker-chosen triggers. Our defense is based on the intuition that the feature extraction layers of a backdoored network embed new features to detect the presence of a trigger and the subsequent classification layers learn to mispredict when triggers are detected. Therefore, to detect backdoors, the proposed defense uses two synergistic anomaly detectors trained on clean validation data: the first is a novelty detector that checks for anomalous features, while the second detects anomalous mappings from features to outputs by comparing with a separate classifier trained on validation data. The approach is evaluated on a wide range of backdoored networks (with multiple variations of triggers) that successfully evade state-of-the-art defenses. Additionally, we evaluate the robustness of our approach on imperceptible perturbations, scalability on large-scale datasets, and effectiveness under domain shift. This paper also shows that the defense can be further improved using data augmentation.
Idiap Research Institute（伊迪亚普研究所）；École Polytéchnique Fédérale de Lausanne (EPFL)（洛桑联邦理工学院（EPFL））
Backdoor attacks allow an attacker to embed functionality jeopardizing proper behavior of any algorithm, machine learning or not. This hidden functionality can remain inactive for normal use of the algorithm until activated by the attacker. Given how stealthy backdoor attacks are, consequences of these backdoors could be disastrous if such networks were to be deployed for applications as critical as border or access control. In this paper, we propose a novel backdoored network detection method based on the principle of anomaly detection, involving access to the clean part of the training data …
No Trust Issues Here: A Technical Report on the Winning Solutions for the Rayan AI Contest
Department of Computer Engineering（计算机工程系）；Bu-Ali Sina University（布阿利·辛纳大学）；Department of Computer Science（计算机科学系）；Iran University of Science and Technology（伊朗科学技术大学）；Technical University of Darmstadt（达姆施塔特技术大学）
System states that are anomalous from the perspective of a domain expert occur frequently in some anomaly detection problems. The performance of commonly used unsupervised anomaly detection methods may suffer in that setting, because they use frequency as a proxy for anomaly. We propose a novel concept for anomaly detection, called relative anomaly detection. It is tailored to be robust towards anomalies that occur frequently, by taking into account their location relative to the most typical observations. The approaches we develop are computationally feasible even for large data sets, and the…
Detecting anomalies in multivariate time-series data is essential in many real-world applications. Recently, various deep learning-based approaches have shown considerable improvements in time-series anomaly detection. However, existing methods still have several limitations, such as long training time due to their complex model designs or costly tuning procedures to find optimal hyperparameters (e.g., sliding window length) for a given dataset. In our paper, we propose a novel method called Implicit Neural Representation-based Anomaly Detection (INRAD). Specifically, we train a simple multi-l…
Detecting faults in manufacturing applications can be difficult, especially if each fault model is to be engineered by hand. Data-driven approaches, using Machine Learning (ML) for detecting faults have recently gained increasing interest, where a ML model can be trained on a set of data from a manufacturing process. In this paper, we present a use case of using ML models for detecting faults during automated screwdriving operations, and introduce a new dataset containing fully monitored and registered data from a Universal Robot and OnRobot screwdriver during both normal and anomalous operati…
School of Electrical Engineering and Computer Science, The Pennsylvania State University（宾夕法尼亚州立大学电气工程与计算机科学学院）；Anomalee Inc.（Anomalee 公司）
Recently, a special type of data poisoning (DP) attack targeting Deep Neural Network (DNN) classifiers, known as a backdoor, was proposed. These attacks do not seek to degrade classification accuracy, but rather to have the classifier learn to classify to a target class whenever the backdoor pattern is present in a test example. Launching backdoor attacks does not require knowledge of the classifier or its training process - it only needs the ability to poison the training set with (a sufficient number of) exemplars containing a sufficiently strong backdoor pattern (labeled with the target cla…
Anomaly detection is a crucial and challenging subject that has been studied within diverse research areas. In this work, we explore the task of log anomaly detection (especially computer system logs and user behavior logs) by analyzing logs' sequential information. We propose LAMA, a multi-head attention based sequential model to process log streams as template activity (event) sequences. A next event prediction task is applied to train the model for anomaly detection. Extensive empirical studies demonstrate that our new model outperforms existing log anomaly detection methods including stati…
NASA Jet Propulsion Laboratory（美国国家航空航天局喷气推进实验室）；California Institute of Technology（加州理工学院）
As spacecraft send back increasing amounts of telemetry data, improved anomaly detection systems are needed to lessen the monitoring burden placed on operations engineers and reduce operational risk. Current spacecraft monitoring systems only target a subset of anomaly types and often require costly expert knowledge to develop and maintain due to challenges involving scale and complexity. We demonstrate the effectiveness of Long Short-Term Memory (LSTMs) networks, a type of Recurrent Neural Network (RNN), in overcoming these issues using expert-labeled telemetry anomaly data from the Soil Mois…
Detecting anomalies and the corresponding root causes in multivariate time series plays an important role in monitoring the behaviors of various real-world systems, e.g., IT system operations or manufacturing industry. Previous anomaly detection approaches model the joint distribution without considering the underlying mechanism of multivariate time series, making them computationally hungry and hard to identify root causes. In this paper, we formulate the anomaly detection problem from a causal perspective and view anomalies as instances that do not follow the regular causal mechanism to gene…
General Electric Global Research Center（通用电气全球研究中心）；General Electric Power & Water Engineering（通用电气电力与水务工程部门）
Monitoring gas turbine combustors health, in particular, early detecting abnormal behaviors and incipient faults, is critical in ensuring gas turbines operating efficiently and in preventing costly unplanned maintenance. One popular means of detecting combustor abnormalities is through continuously monitoring exhaust gas temperature profiles. Over the years many anomaly detection technologies have been explored for detecting combustor faults, however, the performance (detection rate) of anomaly detection solutions fielded is still inadequate. Advanced technologies that can improve detection pe…
George Mason University（乔治梅森大学）；Sandia National Laboratories（桑迪亚国家实验室）
Time series anomaly detection is an important task, with applications in a broad variety of domains. Many approaches have been proposed in recent years, but often they require that the length of the anomalies be known in advance and provided as an input parameter. This limits the practicality of the algorithms, as such information is often unknown in advance, or anomalies with different lengths might co-exist in the data. To address this limitation, previously, a linear time anomaly detection algorithm based on grammar induction has been proposed. While the algorithm can find variable-length p…
Friedrich Schiller University Jena（耶拿弗里德里希·席勒大学）；Carl Zeiss AG（卡尔·蔡司股份公司）
Automatic detection of anomalies in space- and time-varying measurements is an important tool in several fields, e.g., fraud detection, climate analysis, or healthcare monitoring. We present an algorithm for detecting anomalous regions in multivariate spatio-temporal time-series, which allows for spotting the interesting parts in large amounts of data, including video and text data. In opposition to existing techniques for detecting isolated anomalous data points, we propose the "Maximally Divergent Intervals" (MDI) framework for unsupervised detection of coherent spatial regions and time inte…
Inserting a backdoor into the joint model in federated learning (FL) is a recent threat raising concerns. Existing studies mostly focus on developing effective countermeasures against this threat, assuming that backdoored local models, if any, somehow reveal themselves by anomalies in their gradients. However, this assumption needs to be elaborated by identifying specifically which gradients are more likely to indicate an anomaly to what extent under which conditions. This is an important issue given that neural network models usually have huge parametric space and consist of a large number of…
Shanghai Jiao Tong University（上海交通大学）；University of California, Los Angeles（加利福尼亚大学洛杉矶分校）
Deep neural networks have achieved state-of-the-art performance on various tasks. However, lack of interpretability and transparency makes it easier for malicious attackers to inject trojan backdoor into the neural networks, which will make the model behave abnormally when a backdoor sample with a specific trigger is input. In this paper, we propose NeuronInspect, a framework to detect trojan backdoors in deep neural networks via output explanation techniques. NeuronInspect first identifies the existence of backdoor attack targets by generating the explanation heatmap of the output layer. We o…
Outlier detection and novelty detection are two important topics for anomaly detection. Suppose the majority of a dataset are drawn from a certain distribution, outlier detection and novelty detection both aim to detect data samples that do not fit the distribution. Outliers refer to data samples within this dataset, while novelties refer to new samples. In the meantime, backdoor poisoning attacks for machine learning models are achieved through injecting poisoning samples into the training dataset, which could be regarded as "outliers" that are intentionally added by attackers. Differential p…
Deep neural networks are vulnerable to adversarial attacks, such as backdoor attacks in which a malicious adversary compromises a model during training such that specific behaviour can be triggered at test time by attaching a specific word or phrase to an input. This paper considers the problem of diagnosing whether a model has been compromised and if so, identifying the backdoor trigger. We present the first robust defence mechanism that generalizes to several backdoor attacks against text classification models, without prior knowledge of the attack type, nor does our method require access to…
Pennsylvania State University（宾夕法尼亚州立大学）；Shanghai Jiao Tong University（上海交通大学）
Backdoor attacks (BA) are an emerging threat to deep neural network classifiers. A classifier being attacked will predict to the attacker's target class when a test sample from a source class is embedded with the backdoor pattern (BP). Recently, the first BA against point cloud (PC) classifiers was proposed, creating new threats to many important applications including autonomous driving. Such PC BAs are not detectable by existing BA defenses due to their special BP embedding mechanism. In this paper, we propose a reverse-engineering defense that infers whether a PC classifier is backdoor atta…
Predicitions made by neural networks can be fraudulently altered by so-called poisoning attacks. A special case are backdoor poisoning attacks. We study suitable detection methods and introduce a new method called Heatmap Clustering. There, we apply a $k$-means clustering algorithm on heatmaps produced by the state-of-the-art explainable AI method Layer-wise relevance propagation. The goal is to separate poisoned from un-poisoned data in the dataset. We compare this method with a similar method, called Activation Clustering, which also uses $k$-means clustering but applies it on the activation…
While machine learning (ML) models are being increasingly trusted to make decisions in different and varying areas, the safety of systems using such models has become an increasing concern. In particular, ML models are often trained on data from potentially untrustworthy sources, providing adversaries with the opportunity to manipulate them by inserting carefully crafted samples into the training set. Recent work has shown that this type of attack, called a poisoning attack, allows adversaries to insert backdoors or trojans into the model, enabling malicious behavior with simple external backd…
King Abdullah University of Science and Technology (KAUST)（阿卜杜拉国王科技大学）；University of Oxford（牛津大学）
In this paper we investigate the frequency sensitivity of Deep Neural Networks (DNNs) when presented with clean samples versus poisoned samples. Our analysis shows significant disparities in frequency sensitivity between these two types of samples. Building on these findings, we propose FREAK, a frequency-based poisoned sample detection algorithm that is simple yet effective. Our experimental results demonstrate the efficacy of FREAK not only against frequency backdoor attacks but also against some spatial attacks. Our work is just the first step in leveraging these insights. We believe that o…
Deep learning models have recently shown to be vulnerable to backdoor poisoning, an insidious attack where the victim model predicts clean images correctly but classifies the same images as the target class when a trigger poison pattern is added. This poison pattern can be embedded in the training dataset by the adversary. Existing defenses are effective under certain conditions such as a small size of the poison pattern, knowledge about the ratio of poisoned training samples or when a validated clean dataset is available. Since a defender may not have such prior knowledge or resources, we pro…
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）
DMGNN: Detecting and Mitigating Backdoor Attacks in Graph Neural Networks
Nanjing University of Aeronautics and Astronautics（南京航空航天大学）；Yangzhou University（扬州大学）；University of Southern Queensland（南昆士兰大学）；Data61, CSIRO（联邦科学与工业研究组织Data61）；University of Technology Sydney（悉尼科技大学）
Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistency
Huazhong University of Science and Technology（华中科技大学）；National Engineering Research Center for Big Data Technology and System（大数据技术与系统国家工程研究中心）；Wuhan University（武汉大学）
Deep Neural Networks are well known to be vulnerable to adversarial attacks and backdoor attacks, where minor modifications on the input are able to mislead the models to give wrong results. Although defenses against adversarial attacks have been widely studied, investigation on mitigating backdoor attacks is still at an early stage. It is unknown whether there are any connections and common characteristics between the defenses against these two attacks. We conduct comprehensive studies on the connections between adversarial examples and backdoor examples of Deep Neural Networks to seek to ans…
The University of Texas at Dallas（德克萨斯大学达拉斯分校）；The Pennsylvania State University（宾夕法尼亚州立大学）
The black-box nature of deep neural networks (DNNs) facilitates attackers to manipulate the behavior of DNN through data poisoning. Being able to detect and mitigate poisoning attacks, typically categorized into backdoor and adversarial poisoning (AP), is critical in enabling safe adoption of DNNs in many application domains. Although recent works demonstrate encouraging results on detection of certain backdoor attacks, they exhibit inherent limitations which may significantly constrain the applicability. Indeed, no technique exists for detecting AP attacks, which represents a harder challenge…
Defending the Edge: Representative-Attention Defense against Backdoor Attacks in Federated Learning
Dept. of Computer Science Faculty of Engineering University of Manchester Manchester, United Kingdom；Dept. of Computer Science Faculty of Engineering University of Manchester \& MBZUAI Abu Dhabi, United Arab Emirates；Dept. of Computer Science Faculty of Engineering University of Manchester \& KU Leuven Manchester, United Kingdom
Proactive Adversarial Defense: Harnessing Prompt Tuning in Vision-Language Models to Detect Unseen Backdoored Images
University of West Florida（西佛罗里达大学）；Florida Institute For Human and Machine Cognition (IHMC)（佛罗里达人机认知研究所）；Center for Cybersecurity, University of West Florida（西佛罗里达大学网络安全中心）
Shortcuts Everywhere and Nowhere: Exploring Multi-Trigger Backdoor Attacks
Shanghai Key Lab of Intell. Info. Processing, School of CS, Fudan University（上海智能信息处理关键实验室，复旦大学计算机学院）；Bosch Research Asia Pacific & Bosch Center for Artificial Intelligence (BCAI)（博世亚太研究部及博世人工智能中心）；School of Computing and Information Systems, the University of Melbourne（墨尔本大学计算机与信息学院）；School of Computing and Information Systems, Singapore Management University（新加坡管理大学计算机与信息学院）
Neural networks have achieved state-of-the-art performance in solving many problems, including many applications in safety/security-critical systems. Researchers also discovered multiple security issues associated with neural networks. One of them is backdoor attacks, i.e., a neural network may be embedded with a backdoor such that a target output is almost always generated in the presence of a trigger. Existing defense approaches mostly focus on detecting whether a neural network is 'backdoored' based on heuristics, e.g., activation patterns. To the best of our knowledge, the only line of wor…
Backdoor (Trojan) attacks are emerging threats against deep neural networks (DNN). A DNN being attacked will predict to an attacker-desired target class whenever a test sample from any source class is embedded with a backdoor pattern; while correctly classifying clean (attack-free) test samples. Existing backdoor defenses have shown success in detecting whether a DNN is attacked and in reverse-engineering the backdoor pattern in a "post-training" regime: the defender has access to the DNN to be inspected and a small, clean dataset collected independently, but has no access to the (possibly poi…
Deep neural networks (DNNs) are proved to be vulnerable against backdoor attacks. A backdoor is often embedded in the target DNNs through injecting a backdoor trigger into training examples, which can cause the target DNNs misclassify an input attached with the backdoor trigger. Existing backdoor detection methods often require the access to the original poisoned training data, the parameters of the target DNNs, or the predictive confidence for each given input, which are impractical in many real-world applications, e.g., on-device deployed DNNs. We address the black-box hard-label backdoor de…
Xidian University（西安电子科技大学）；University of Copenhagen（哥本哈根大学）；Sony AI（索尼人工智能）；University of Illinois at Urbana–Champaign（伊利诺伊大学厄巴纳-香槟分校）；Deakin University（迪肯大学）
Backdoor attack has emerged as a major security threat to deep neural networks (DNNs). While existing defense methods have demonstrated promising results on detecting or erasing backdoors, it is still not clear whether robust training methods can be devised to prevent the backdoor triggers being injected into the trained model in the first place. In this paper, we introduce the concept of \emph{anti-backdoor learning}, aiming to train \emph{clean} models given backdoor-poisoned data. We frame the overall learning process as a dual-task of learning the \emph{clean} and the \emph{backdoor} porti…
