Preventing Collapse in Contrastive Learning with Orthonormal Prototypes (CLOP)
This addresses neural collapse in semi-supervised and self-supervised contrastive learning, which is an incremental improvement over prior methods like simplex ETF structures.
The paper tackles the problem of neural collapse in contrastive learning, where embeddings converge into a lower-dimensional space, by proposing CLOP, a novel semi-supervised loss function that promotes orthogonal linear subspaces among class embeddings. The result shows that CLOP enhances performance, providing greater stability across different learning rates and batch sizes in experiments on real and synthetic datasets.
Contrastive learning has emerged as a powerful method in deep learning, excelling at learning effective representations through contrasting samples from different distributions. However, neural collapse, where embeddings converge into a lower-dimensional space, poses a significant challenge, especially in semi-supervised and self-supervised setups. In this paper, we first theoretically analyze the effect of large learning rates on contrastive losses that solely rely on the cosine similarity metric, and derive a theoretical bound to mitigate this collapse. {Building on these insights, we propose CLOP, a novel semi-supervised loss function designed to prevent neural collapse by promoting the formation of orthogonal linear subspaces among class embeddings.} Unlike prior approaches that enforce a simplex ETF structure, CLOP focuses on subspace separation, leading to more distinguishable embeddings. Through extensive experiments on real and synthetic datasets, we demonstrate that CLOP enhances performance, providing greater stability across different learning rates and batch sizes.