Generalization Bounds for Robust Contrastive Learning: From Theory to Practice
This work addresses a theoretical gap in robust contrastive learning for machine learning practitioners, but it is incremental as it builds on existing ACL methods.
The paper tackles the lack of theoretical understanding in Adversarial Contrastive Learning (ACL) by developing rigorous theories to identify which unsupervised training components improve robust supervised loss, revealing that adversarial contrastive loss, benign loss, and a global divergence between benign and adversarial examples contribute to robustness.
Contrastive Learning first extracts features from unlabeled data, followed by linear probing with labeled data. Adversarial Contrastive Learning (ACL) integrates Adversarial Training into the first phase to enhance feature robustness against attacks in the probing phase. While ACL has shown strong empirical results, its theoretical understanding remains limited. Furthermore, while a fair amount of theoretical works analyze how the unsupervised loss can support the supervised loss in the probing phase, none has examined its role to the robust supervised loss. To fill this gap, our work develops rigorous theories to identify which components in the unsupervised training can help improve the robust supervised loss. Specifically, besides the adversarial contrastive loss, we reveal that the benign one, along with a global divergence between benign and adversarial examples can also improve robustness. Proper experiments are conducted to justify our findings.