اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدDefending against Backdoor Attacks in Natural Language Generation
436
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
The frustratingly fragile nature of neural network models make current natural language generation (NLG) systems prone to backdoor attacks and generate malicious sequences that could be sexist or offensive. Unfortunately, little effort has been invested to how backdoor attacks can affect current NLG models and how to defend against these attacks. In this work, we investigate this problem on two important NLG tasks, machine translation and dialogue generation. By giving a formal definition for backdoor attack and defense, and developing corresponding benchmarks, we design methods to attack NLG models, which achieve high attack success to ask NLG models to generate malicious sequences. To defend against these attacks, we propose to detect the attack trigger by examining the effect of deleting or replacing certain words on the generation outputs, which we find successful for certain types of attacks. We will discuss the limitation of this work, and hope this work can raise the awareness of backdoor risks concealed in deep NLG systems. (Code and data are available at https://github.com/ShannonAI/backdoor_nlg.)
قيم البحث
اقرأ أيضاً
Modern machine learning increasingly requires training on a large collection of data from multiple sources, not all of which can be trusted. A particularly concerning scenario is when a small fraction of poisoned data changes the behavior of the trai
ned model when triggered by an attacker-specified watermark. Such a compromised model will be deployed unnoticed as the model is accurate otherwise. There have been promising attempts to use the intermediate representations of such a model to separate corrupted examples from clean ones. However, these defenses work only when a certain spectral signature of the poisoned examples is large enough for detection. There is a wide range of attacks that cannot be protected against by the existing defenses. We propose a novel defense algorithm using robust covariance estimation to amplify the spectral signature of corrupted data. This defense provides a clean model, completely removing the backdoor, even in regimes where previous methods have no hope of detecting the poisoned examples. Code and pre-trained models are available at https://github.com/SewoongLab/spectre-defense .
Although deep neural networks (DNNs) have achieved a great success in various computer vision tasks, it is recently found that they are vulnerable to adversarial attacks. In this paper, we focus on the so-called textit{backdoor attack}, which injects
a backdoor trigger to a small portion of training data (also known as data poisoning) such that the trained DNN induces misclassification while facing examples with this trigger. To be specific, we carefully study the effect of both real and synthetic backdoor attacks on the internal response of vanilla and backdoored DNNs through the lens of Gard-CAM. Moreover, we show that the backdoor attack induces a significant bias in neuron activation in terms of the $ell_infty$ norm of an activation map compared to its $ell_1$ and $ell_2$ norm. Spurred by our results, we propose the textit{$ell_infty$-based neuron pruning} to remove the backdoor from the backdoored DNN. Experiments show that our method could effectively decrease the attack success rate, and also hold a high classification accuracy for clean images.
Recent studies have shown that deep neural networks (DNNs) are highly vulnerable to adversarial attacks, including evasion and backdoor (poisoning) attacks. On the defense side, there have been intensive interests in both empirical and provable robus
tness against evasion attacks; however, provable robustness against backdoor attacks remains largely unexplored. In this paper, we focus on certifying robustness against backdoor attacks. To this end, we first provide a unified framework for robustness certification and show that it leads to a tight robustness condition for backdoor attacks. We then propose the first robust training process, RAB, to smooth the trained model and certify its robustness against backdoor attacks. Moreover, we evaluate the certified robustness of a family of smoothed models which are trained in a differentially private fashion, and show that they achieve better certified robustness bounds. In addition, we theoretically show that it is possible to train the robust smoothed models efficiently for simple models such as K-nearest neighbor classifiers, and we propose an exact smooth-training algorithm which eliminates the need to sample from a noise distribution. Empirically, we conduct comprehensive experiments for different machine learning (ML) models such as DNNs, differentially private DNNs, and K-NN models on MNIST, CIFAR-10 and ImageNet datasets (focusing on binary classifiers), and provide the first benchmark for certified robustness against backdoor attacks. In addition, we evaluate K-NN models on a spambase tabular dataset to demonstrate the advantages of the proposed exact algorithm. Both the theoretical analysis and the comprehensive benchmark on diverse ML models and datasets shed lights on further robust learning strategies against training time attacks or other general adversarial attacks.
Machine learning (ML) has progressed rapidly during the past decade and ML models have been deployed in various real-world applications. Meanwhile, machine learning models have been shown to be vulnerable to various security and privacy attacks. One
attack that has attracted a great deal of attention recently is the backdoor attack. Specifically, the adversary poisons the target model training set, to mislead any input with an added secret trigger to a target class, while keeping the accuracy for original inputs unchanged. Previous backdoor attacks mainly focus on computer vision tasks. In this paper, we present the first systematic investigation of the backdoor attack against models designed for natural language processing (NLP) tasks. Specifically, we propose three methods to construct triggers in the NLP setting, including Char-level, Word-level, and Sentence-level triggers. Our Attacks achieve an almost perfect success rate without jeopardizing the original model utility. For instance, using the word-level triggers, our backdoor attack achieves 100% backdoor accuracy with only a drop of 0.18%, 1.26%, and 0.19% in the models utility, for the IMDB, Amazon, and Stanford Sentiment Treebank datasets, respectively.
Robustness against word substitutions has a well-defined and widely acceptable form, i.e., using semantically similar words as substitutions, and thus it is considered as a fundamental stepping-stone towards broader robustness in natural language pro
cessing. Previous defense methods capture word substitutions in vector space by using either $l_2$-ball or hyper-rectangle, which results in perturbation sets that are not inclusive enough or unnecessarily large, and thus impedes mimicry of worst cases for robust training. In this paper, we introduce a novel textit{Adversarial Sparse Convex Combination} (ASCC) method. We model the word substitution attack space as a convex hull and leverages a regularization term to enforce perturbation towards an actual substitution, thus aligning our modeling better with the discrete textual space. Based on the ASCC method, we further propose ASCC-defense, which leverages ASCC to generate worst-case perturbations and incorporates adversarial training towards robustness. Experiments show that ASCC-defense outperforms the current state-of-the-arts in terms of robustness on two prevailing NLP tasks, emph{i.e.}, sentiment analysis and natural language inference, concerning several attacks across multiple model architectures. Besides, we also envision a new class of defense towards robustness in NLP, where our robustly trained word vectors can be plugged into a normally trained model and enforce its robustness without applying any other defense techniques.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
