InfoNCE: Identifying the Gap Between Theory and Practice
This work addresses a foundational problem in machine learning by bridging theory and practice for researchers, but it is incremental as it builds on existing InfoNCE frameworks.
The paper tackles the gap between theoretical assumptions and practical implementations in contrastive learning with InfoNCE loss, showing that a more realistic anisotropic setting allows for better recovery of latent factors, with improvements like increased information recovery on CIFAR10 and ImageNet, though at a cost to downstream accuracy.
Prior theory work on Contrastive Learning via the InfoNCE loss showed that, under certain assumptions, the learned representations recover the ground-truth latent factors. We argue that these theories overlook crucial aspects of how CL is deployed in practice. Specifically, they either assume equal variance across all latents or that certain latents are kept invariant. However, in practice, positive pairs are often generated using augmentations such as strong cropping to just a few pixels. Hence, a more realistic assumption is that all latent factors change with a continuum of variability across all factors. We introduce AnInfoNCE, a generalization of InfoNCE that can provably uncover the latent factors in this anisotropic setting, broadly generalizing previous identifiability results in CL. We validate our identifiability results in controlled experiments and show that AnInfoNCE increases the recovery of previously collapsed information in CIFAR10 and ImageNet, albeit at the cost of downstream accuracy. Finally, we discuss the remaining mismatches between theoretical assumptions and practical implementations.