Towards Understanding Fast Adversarial Training
This work addresses the scalability problem for machine learning practitioners by making adversarial defenses more efficient, though it is incremental as it builds on existing fast adversarial training methods.
The paper tackled the computational expense of adversarial training by analyzing fast adversarial training, showing its success stems from recovering from overfitting to weak attacks, and improved it to achieve superior robust accuracy with reduced training time compared to strong adversarial training.
Current neural-network-based classifiers are susceptible to adversarial examples. The most empirically successful approach to defending against such adversarial examples is adversarial training, which incorporates a strong self-attack during training to enhance its robustness. This approach, however, is computationally expensive and hence is hard to scale up. A recent work, called fast adversarial training, has shown that it is possible to markedly reduce computation time without sacrificing significant performance. This approach incorporates simple self-attacks, yet it can only run for a limited number of training epochs, resulting in sub-optimal performance. In this paper, we conduct experiments to understand the behavior of fast adversarial training and show the key to its success is the ability to recover from overfitting to weak attacks. We then extend our findings to improve fast adversarial training, demonstrating superior robust accuracy to strong adversarial training, with much-reduced training time.