Model Compression with Adversarial Robustness: A Unified Optimization Framework
This work addresses the challenge of maintaining adversarial robustness in compressed models, which is important for deploying efficient and secure AI systems, representing an incremental improvement by combining existing compression means with robustness constraints.
The paper tackles the problem of compressing deep models without compromising their adversarial robustness, proposing a unified optimization framework that integrates various compression techniques and achieves a more favorable trade-off among model size, accuracy, and robustness compared to existing methods.
Deep model compression has been extensively studied, and state-of-the-art methods can now achieve high compression ratios with minimal accuracy loss. This paper studies model compression through a different lens: could we compress models without hurting their robustness to adversarial attacks, in addition to maintaining accuracy? Previous literature suggested that the goals of robustness and compactness might sometimes contradict. We propose a novel Adversarially Trained Model Compression (ATMC) framework. ATMC constructs a unified constrained optimization formulation, where existing compression means (pruning, factorization, quantization) are all integrated into the constraints. An efficient algorithm is then developed. An extensive group of experiments are presented, demonstrating that ATMC obtains remarkably more favorable trade-off among model size, accuracy and robustness, over currently available alternatives in various settings. The codes are publicly available at: https://github.com/shupenggui/ATMC.