Yaxin Li

Shenglai Zeng, Yaxin Li, Jie Ren et al.

Large language models (LLMs) have shown great capabilities in various tasks but also exhibited memorization of training data, raising tremendous privacy and copyright concerns. While prior works have studied memorization during pre-training, the exploration of memorization during fine-tuning is rather limited. Compared to pre-training, fine-tuning typically involves more sensitive data and diverse objectives, thus may bring distinct privacy risks and unique memorization behaviors. In this work, we conduct the first comprehensive analysis to explore language models' (LMs) memorization during fine-tuning across tasks. Our studies with open-sourced and our own fine-tuned LMs across various tasks indicate that memorization presents a strong disparity among different fine-tuning tasks. We provide an intuitive explanation of this task disparity via sparse coding theory and unveil a strong correlation between memorization and attention score distribution.

Yaxin Li, Xiaorui Liu, Han Xu et al.

Deep Neural Network (DNN) are vulnerable to adversarial attacks. As a countermeasure, adversarial training aims to achieve robustness based on the min-max optimization problem and it has shown to be one of the most effective defense strategies. However, in this work, we found that compared with natural training, adversarial training fails to learn better feature representations for either clean or adversarial samples, which can be one reason why adversarial training tends to have severe overfitting issues and less satisfied generalize performance. Specifically, we observe two major shortcomings of the features learned by existing adversarial training methods:(1) low intra-class feature similarity; and (2) conservative inter-classes feature variance. To overcome these shortcomings, we introduce a new concept of adversarial training graph (ATG) with which the proposed adversarial training with feature separability (ATFS) enables to coherently boost the intra-class feature similarity and increase inter-class feature variance. Through comprehensive experiments, we demonstrate that the proposed ATFS framework significantly improves both clean and robust performance.

Jie Ren, Yaxin Li, Shenglai Zeng et al.

Recent advancements in text-to-image diffusion models have demonstrated their remarkable capability to generate high-quality images from textual prompts. However, increasing research indicates that these models memorize and replicate images from their training data, raising tremendous concerns about potential copyright infringement and privacy risks. In our study, we provide a novel perspective to understand this memorization phenomenon by examining its relationship with cross-attention mechanisms. We reveal that during memorization, the cross-attention tends to focus disproportionately on the embeddings of specific tokens. The diffusion model is overfitted to these token embeddings, memorizing corresponding training images. To elucidate this phenomenon, we further identify and discuss various intrinsic findings of cross-attention that contribute to memorization. Building on these insights, we introduce an innovative approach to detect and mitigate memorization in diffusion models. The advantage of our proposed method is that it will not compromise the speed of either the training or the inference processes in these models while preserving the quality of generated images. Our code is available at https://github.com/renjie3/MemAttn .

Xiaorui Liu, Wei Jin, Yao Ma et al.

While many existing graph neural networks (GNNs) have been proven to perform $\ell_2$-based graph smoothing that enforces smoothness globally, in this work we aim to further enhance the local smoothness adaptivity of GNNs via $\ell_1$-based graph smoothing. As a result, we introduce a family of GNNs (Elastic GNNs) based on $\ell_1$ and $\ell_2$-based graph smoothing. In particular, we propose a novel and general message passing scheme into GNNs. This message passing algorithm is not only friendly to back-propagation training but also achieves the desired smoothing properties with a theoretical convergence guarantee. Experiments on semi-supervised learning tasks demonstrate that the proposed Elastic GNNs obtain better adaptivity on benchmark datasets and are significantly robust to graph adversarial attacks. The implementation of Elastic GNNs is available at \url{https://github.com/lxiaorui/ElasticGNN}.

Yaxin Li, Wei Jin, Han Xu et al.

DeepRobust is a PyTorch adversarial learning library which aims to build a comprehensive and easy-to-use platform to foster this research field. It currently contains more than 10 attack algorithms and 8 defense algorithms in image domain and 9 attack algorithms and 4 defense algorithms in graph domain, under a variety of deep learning architectures. In this manual, we introduce the main contents of DeepRobust with detailed instructions. The library is kept updated and can be found at https://github.com/DSE-MSU/DeepRobust.

Wei Jin, Yaxin Li, Han Xu et al.

Deep neural networks (DNNs) have achieved significant performance in various tasks. However, recent studies have shown that DNNs can be easily fooled by small perturbation on the input, called adversarial attacks. As the extensions of DNNs to graphs, Graph Neural Networks (GNNs) have been demonstrated to inherit this vulnerability. Adversary can mislead GNNs to give wrong predictions by modifying the graph structure such as manipulating a few edges. This vulnerability has arisen tremendous concerns for adapting GNNs in safety-critical applications and has attracted increasing research attention in recent years. Thus, it is necessary and timely to provide a comprehensive overview of existing graph adversarial attacks and the countermeasures. In this survey, we categorize existing attacks and defenses, and review the corresponding state-of-the-art methods. Furthermore, we have developed a repository with representative algorithms (https://github.com/DSE-MSU/DeepRobust/tree/master/deeprobust/graph). The repository enables us to conduct empirical studies to deepen our understandings on attacks and defenses on graphs.

Wentao Wang, Han Xu, Xiaorui Liu et al.

Adversarial training has been empirically proven to be one of the most effective and reliable defense methods against adversarial attacks. However, almost all existing studies about adversarial training are focused on balanced datasets, where each class has an equal amount of training examples. Research on adversarial training with imbalanced training datasets is rather limited. As the initial effort to investigate this problem, we reveal the facts that adversarially trained models present two distinguished behaviors from naturally trained models in imbalanced datasets: (1) Compared to natural training, adversarially trained models can suffer much worse performance on under-represented classes, when the training dataset is extremely imbalanced. (2) Traditional reweighting strategies may lose efficacy to deal with the imbalance issue for adversarial training. For example, upweighting the under-represented classes will drastically hurt the model's performance on well-represented classes, and as a result, finding an optimal reweighting value can be tremendously challenging. In this paper, to further understand our observations, we theoretically show that the poor data separability is one key reason causing this strong tension between under-represented and well-represented classes. Motivated by this finding, we propose Separable Reweighted Adversarial Training (SRAT) to facilitate adversarial training under imbalanced scenarios, by learning more separable features for different classes. Extensive experiments on various datasets verify the effectiveness of the proposed framework.

Han Xu, Yaxin Li, Xiaorui Liu et al.

Meta learning algorithms have been widely applied in many tasks for efficient learning, such as few-shot image classification and fast reinforcement learning. During meta training, the meta learner develops a common learning strategy, or experience, from a variety of learning tasks. Therefore, during meta test, the meta learner can use the learned strategy to quickly adapt to new tasks even with a few training samples. However, there is still a dark side about meta learning in terms of reliability and robustness. In particular, is meta learning vulnerable to adversarial attacks? In other words, would a well-trained meta learner utilize its learned experience to build wrong or likely useless knowledge, if an adversary unnoticeably manipulates the given training set? Without the understanding of this problem, it is extremely risky to apply meta learning in safety-critical applications. Thus, in this paper, we perform the initial study about adversarial attacks on meta learning under the few-shot classification problem. In particular, we formally define key elements of adversarial attacks unique to meta learning and propose the first attacking algorithm against meta learning under various settings. We evaluate the effectiveness of the proposed attacking strategy as well as the robustness of several representative meta learning algorithms. Experimental results demonstrate that the proposed attacking strategy can easily break the meta learner and meta learning is vulnerable to adversarial attacks. The implementation of the proposed framework will be released upon the acceptance of this paper.