Sentiment classification has been broadly applied in real life, such as product recommendation and opinion-oriented analysis. Unfortunately, the widely employed sentiment classification systems based on deep neural networks (DNNs) are susceptible to adversarial attacks with imperceptible perturbations into the legitimate texts (also called adversarial texts ). Adversarial texts could cause erroneous outputs even without access to the target model, bringing security concerns to systems deployed in safety-critical applications. However, studies on defending against adversarial texts are still in the early stage and not ready for tackling the emerging threats, especially in dealing with unknown attacks. Investigating the minor differences between adversarial texts and legitimate texts and enhancing the robustness of target models are two mainstream ideas for defending against adversarial texts. However, both of them suffer the generalization issue in dealing with unknown adversarial attacks. In this paper, we proposed a general method, called TextFirewall , for defending against adversarial texts crafted by various adversarial attacks, which shows the potential in identifying new developed adversarial attacks in the future. Given a piece of text, our TextFirewall identifies the adversarial text by investigating the inconsistency between the target model’s output and the impact value calculated by important words in the text. TextFirewall could be deployed as a third-party tool without modifying the target model and agnostic to the specific type of adversarial texts. Experimental results demonstrate that our proposed TextFirewall effectively identifies adversarial texts generated by the three state-of-the-art (SOTA) attacks and outperforms previous defense techniques. Specifically, TextFirewall achieves an average accuracy of 90.7% on IMDB and 96.9% on Yelp in defending the three SOTA attacks.
Watermarking has been widely adopted for protecting the intellectual property (IP) of Deep Neural Networks (DNN) to defend the unauthorized distribution. Unfortunately, studies have shown that the popular data-poisoning DNN watermarking scheme via tedious model fine-tuning on a poisoned dataset (carefully-crafted sample-label pairs) is not efficient in tackling the tasks on challenging datasets and production-level DNN model protection. To address the aforementioned limitation, in this paper, we propose a plug-and-play watermarking scheme for DNN models by injecting an independent proprietary model into the target model to serve the watermark embedding and ownership verification. In contrast to the prior studies, our proposed method by incorporating a proprietary model is free of target model fine-tuning without involving any parameters update of the target model, thus the fidelity is well preserved and scalable to challenging real tasks. Experimental results on real-world challenging datasets (e.g., ImageNet) and production-level DNN models demonstrated its effectiveness, fidelity w.r.t. the functionality preservation of the target model, robustness against popular watermark removal attacks, and the plug-and-play deployment. The source code and models are available at https://github.com/AntigoneRandy/PTYNet.
14 August 2019 in IEEE Transactions on Mobile Computing
With the unprecedented convenience brought by Apps on mobile devices, we are facing severe security attacks and privacy leakage caused by them since they may stealthily access unclaimed or unneeded permissions for some purposes. Many works strive to discover these malicious apps using program analysis techniques, however, they fail to tell users why an app needs to request the permission from users' perspective. In this paper, we leverage the power of the crowdsourced user reviews to understand why an app requests a permission. We propose a framework, called SmartPI, that automatically identifies functionality-relevant user reviews and infers the permission implication of them, bridging the gap between the functionalities and the actual behaviors of an app. In particular, we extract features from the platform documents to identify functionality-relevant user reviews from noisy crowdsourced user reviews with Natural Language Processing (NLP) techniques. The topic model is further adopted to infer the permission implications of apps from the functionality-relevant user reviews. More than 20,000 apps, 2,653,159 users and 4,247,769 user reviews are crawled from Google Play as a real-world dataset to evaluate the performance of SmartPI. The experiments results show that the permission usage of apps can be better reflected by user reviews than the claimed descriptions of apps.
24 March 2024 in Proceedings of the AAAI Conference on Artificial Intelligence
In recent few years, DeepFakes are posing serve threats and concerns to both individuals and celebrities, as realistic DeepFakes facilitate the spread of disinformation. Model attribution techniques aim at attributing the adopted forgery models of DeepFakes for provenance purposes and providing explainable results to DeepFake forensics. However, the existing model attribution techniques rely on the trace left in the DeepFake creation, which can become futile if such traces were disrupted. Motivated by our observation that certain traces served for model attribution appeared in both the high-frequency and low-frequency domains and play a divergent role in model attribution. In this work, for the first time, we propose a novel training-free evasion attack, TraceEvader, in the most practical non-box setting. Specifically, TraceEvader injects a universal imitated traces learned from wild DeepFakes into the high-frequency component and introduces adversarial blur into the domain of the low-frequency component, where the added distortion confuses the extraction of certain traces for model attribution. The comprehensive evaluation on 4 state-of-the-art (SOTA) model attribution techniques and fake images generated by 8 generative models including generative adversarial networks (GANs) and diffusion models (DMs) demonstrates the effectiveness of our method. Overall, our TraceEvader achieves the highest average attack success rate of 79% and is robust against image transformations and dedicated denoising techniques as well where the average attack success rate is still around 75%. Our TraceEvader confirms the limitations of current model attribution techniques and calls the attention of DeepFake researchers and practitioners for more robust-purpose model attribution techniques.
12 December 2024 in IEEE Transactions on Information Forensics and Security
Adversarial training (AT) is widely considered the state-of-the-art technique for improving the robustness of deep neural networks (DNNs) against adversarial examples (AEs). Nevertheless, recent studies have revealed that adversarially trained models are prone to unfairness problems. Recent works in this field usually apply class-wise regularization methods to enhance the fairness of AT. However, this paper discovers that these paradigms can be sub-optimal in improving robust fairness. Specifically, we empirically observe that the AEs that are already robust (referred to as “easy AEs” in this paper) are useless and even harmful in improving robust fairness. To this end, we propose the hard adversarial example mining (HAM) technique which concentrates on mining hard AEs while discarding the easy AEs in AT. Specifically, HAM identifies the easy AEs and hard AEs with a fast adversarial attack method. By discarding the easy AEs and reweighting the hard AEs, the robust fairness of the model can be efficiently and effectively improved. Extensive experimental results on four image classification datasets demonstrate the improvement of HAM in robust fairness and training efficiency compared to several state-of-the-art fair adversarial training methods. Our code will be made publicly available .