An official website of the United States government
Adversarial training (AT) is a widely recognized defense mechanism to gain the robustness of deep neural networks against adversarial attacks. It is built on min-max optimization (MMO), where the minimizer (i.e., defender) seeks a robust model to minimize the worst-case training loss in the presence of adversarial examples crafted by the maximizer (i.e., attacker). However, the conventional MMO method makes AT hard to scale. Thus, FAST-AT (Wong et al., 2020) and other recent algorithms attempt to simplify MMO by replacing its maximization step with the single gradient sign-based attack generation step. Although easy to implement, FAST-AT lacks theoretical guarantees, and its empirical performance is unsatisfactory due to the issue of robust catastrophic overfitting when training with strong adversaries. In this paper, we advance FAST-AT from the fresh perspective of bi-level optimization (BLO). We first show that the commonly used FAST-AT is equivalent to using a stochastic gradient algorithm to solve a linearized BLO problem involving a sign operation. However, the discrete nature of the sign operation makes it difficult to understand the algorithm performance. Inspired by BLO, we design and analyze a new set of robust training algorithms termed Fast Bilevel AT (FAST-BAT), which effectively defends sign-based projected gradient descent (PGD) attacks without using any gradient sign method or explicit robust regularization. In practice, we show our method yields substantial robustness improvements over baselines across multiple models and datasets
more »
« less
Distributed Adversarial Training to Robustify Deep Neural Networks at Scale
Current deep neural networks (DNNs) are vulnerable to adversarial attacks, where adversarial perturbations to the inputs can change or manipulate classification. To defend against such attacks, an effective and popular approach, known as adversarial training (AT), has been shown to mitigate the negative impact of adversarial attacks by virtue of a min-max robust training method. While effective, it remains unclear whether it can successfully be adapted to the distributed learning context. The power of distributed optimization over multiple machines enables us to scale up robust training over large models and datasets. Spurred by that, we propose distributed adversarial training (DAT), a large-batch adversarial training framework implemented over multiple machines. We show that DAT is general, which supports training over labeled and unlabeled data, multiple types of attack generation methods, and gradient compression operations favored for distributed optimization. Theoretically, we provide, under standard conditions in the optimization theory, the convergence rate of DAT to the first-order stationary points in general non-convex settings. Empirically, we demonstrate that DAT either matches or outperforms state-of-the-art robust accuracies and achieves a graceful training speedup (e.g., on ResNet–50 under ImageNet).
more »
« less
- PAR ID:
- 10341715
- Date Published:
- Journal Name:
- Uncertainty in artificial intelligence
- ISSN:
- 1525-3384
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Deep Neural Network (DNN) trained by the gradient descent method is known to be vulnerable to maliciously perturbed adversarial input, aka. adversarial attack. As one of the countermeasures against adversarial attacks, increasing the model capacity for DNN robustness enhancement was discussed and reported as an effective approach by many recent works. In this work, we show that shrinking the model size through proper weight pruning can even be helpful to improve the DNN robustness under adversarial attack. For obtaining a simultaneously robust and compact DNN model, we propose a multi-objective training method called Robust Sparse Regularization (RSR), through the fusion of various regularization techniques, including channel-wise noise injection, lasso weight penalty, and adversarial training. We conduct extensive experiments to show the effectiveness of RSR against popular white-box (i.e., PGD and FGSM) and black-box attacks. Thanks to RSR, 85 % weight connections of ResNet-18 can be pruned while still achieving 0.68 % and 8.72 % improvement in clean- and perturbed-data accuracy respectively on CIFAR-10 dataset, in comparison to its PGD adversarial training baseline.more » « less
-
Recent advances in machine learning enable wider applications of prediction models in cyber-physical systems. Smart grids are increasingly using distributed sensor settings for distributed sensor fusion and information processing. Load forecasting systems use these sensors to predict future loads to incorporate into dynamic pricing of power and grid maintenance. However, these inference predictors are highly complex and thus vulnerable to adversarial attacks. Moreover, the adversarial attacks are synthetic norm-bounded modifications to a limited number of sensors that can greatly affect the accuracy of the overall predictor. It can be much cheaper and effective to incorporate elements of security and resilience at the earliest stages of design. In this paper, we demonstrate how to analyze the security and resilience of learning-based prediction models in power distribution networks by utilizing a domain-specific deep-learning and testing framework. This framework is developed using DeepForge and enables rapid design and analysis of attack scenarios against distributed smart meters in a power distribution network. It runs the attack simulations in the cloud backend. In addition to the predictor model, we have integrated an anomaly detector to detect adversarial attacks targeting the predictor. We formulate the stealthy adversarial attacks as an optimization problem to maximize prediction loss while minimizing the required perturbations. Under the worst-case setting, where the attacker has full knowledge of both the predictor and the detector, an iterative attack method has been developed to solve for the adversarial perturbation. We demonstrate the framework capabilities using a GridLAB-D based power distribution network model and show how stealthy adversarial attacks can affect smart grid prediction systems even with a partial control of network.more » « less
-
Recent advancements in Large Language Models (LLMs) have showcased remarkable capabilities across various tasks in different domains. However, the emergence of biases and the potential for generating harmful content in LLMs, particularly under malicious inputs, pose significant challenges. Current mitigation strategies, while effective, are not resilient under adversarial attacks. This paper introduces Resilient Guardrails for Large Language Models (RigorLLM), a novel framework designed to efficiently and effectively moderate harmful and unsafe inputs and outputs for LLMs. By employing a multi-faceted approach that includes energy-based training data augmentation through Langevin dynamics, optimizing a safe suffix for inputs via minimax optimization, and integrating a fusion-based model combining robust KNN with LLMs based on our data augmentation, RigorLLM offers a robust solution to harmful content moderation. Our experimental evaluations demonstrate that RigorLLM not only outperforms existing baselines like OpenAI API and Perspective API in detecting harmful content but also exhibits unparalleled resilience to jailbreaking attacks. The innovative use of constrained optimization and a fusion-based guardrail approach represents a significant step forward in developing more secure and reliable LLMs, setting a new standard for content moderation frameworks in the face of evolving digital threats.more » « less
-
Adversarial training has emerged as a popular approach for training models that are robust to inference-time adversarial attacks. However, our theoretical understanding of why and when it works remains limited. Prior work has offered generalization analysis of adversarial training, but they are either restricted to the Neural Tangent Kernel (NTK) regime or they make restrictive assumptions about data such as (noisy) linear separability or robust realizability. In this work, we study the stability and generalization of adversarial training for two-layer networks without any data distribution assumptions and beyond the NTK regime. Our findings suggest that for networks with any given initialization and sufficiently large width, the generalization bound can be effectively controlled via early stopping. We further improve the generalization bound by leveraging smoothing using Moreau’s envelope.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
