Revisiting Theory of Contrastive Learning for Domain Generalization
By: Ali Alvandi, Mina Rezaei
Potential Business Impact:
Helps computers learn from new, unseen data.
Contrastive learning is among the most popular and powerful approaches for self-supervised representation learning, where the goal is to map semantically similar samples close together while separating dissimilar ones in the latent space. Existing theoretical methods assume that downstream task classes are drawn from the same latent class distribution used during the pretraining phase. However, in real-world settings, downstream tasks may not only exhibit distributional shifts within the same label space but also introduce new or broader label spaces, leading to domain generalization challenges. In this work, we introduce novel generalization bounds that explicitly account for both types of mismatch: domain shift and domain generalization. Specifically, we analyze scenarios where downstream tasks either (i) draw classes from the same latent class space but with shifted distributions, or (ii) involve new label spaces beyond those seen during pretraining. Our analysis reveals how the performance of contrastively learned representations depends on the statistical discrepancy between pretraining and downstream distributions. This extended perspective allows us to derive provable guarantees on the performance of learned representations on average classification tasks involving class distributions outside the pretraining latent class set.
Similar Papers
Connecting Domains and Contrasting Samples: A Ladder for Domain Generalization
CV and Pattern Recognition
Helps computers learn from different data types.
Effect of Domain Generalization Techniques in Low Resource Systems
Computation and Language
Makes computer language learning work better everywhere.
An Augmentation Overlap Theory of Contrastive Learning
Machine Learning (CS)
Teaches computers to group similar things without labels.