Why Prototypes Collapse: Diagnosing and Preventing Partial Collapse in Prototypical Self-Supervised Learning

TL;DR

This paper identifies the root cause of partial prototype collapse in self-supervised learning and proposes a decoupled training method that maintains diverse prototypes and improves downstream results.

Contribution

It introduces a novel decoupled training strategy that separates prototype and encoder optimization, preventing collapse without extra regularizers.

Findings

Decoupled training prevents prototype collapse effectively.

Diverse prototypes lead to better downstream performance.

The method outperforms existing regularization techniques.

Abstract

Prototypical self-supervised learning methods consistently suffer from partial prototype collapse, where multiple prototypes converge to nearly identical representations. This undermines their central purpose -- providing diverse and informative targets to guide encoders toward rich representations -- and has led practitioners to over-parameterize prototype sets or add ad-hoc regularizers, which mitigate symptoms rather than address the root cause. We empirically trace the collapse to the joint optimization of encoders and prototypes, which encourages a type of shortcut learning: early in training prototypes drift toward redundant representations that minimize loss without necessarily enhancing representation diversity. To break the joint optimization, we introduce a fully decoupled training strategy that learns prototypes and encoders under separate objectives. Concretely, we model prototypes as a Gaussian mixture updated with an online EM-style procedure, independent of the encoder's loss. This simple yet principled decoupling eliminates prototype collapse without explicit regularization and yields consistently diverse prototypes and stronger downstream performance.