The Role of Entropy and Reconstruction in Multi-View Self-Supervised Learning
Borja Rodru0131u0301guez Gu00e1lvez,u00a0Arno Blaas,u00a0Pau Rodriguez,u00a0Adam Golinski,u00a0Xavier Suau,u00a0Jason Ramapuram,u00a0Dan Busbridge,u00a0Luca Zappella
The mechanisms behind the success of multi-view self-supervised learning (MVSSL) are not yet fully understood. Contrastive MVSSL methods have been studied through the lens of InfoNCE, a lower bound of the Mutual Information (MI). However, the relation between other MVSSL methods and MI remains unclear. We consider a different lower bound on the MI consisting of an entropy and a reconstruction term (ER), and analyze the main MVSSL families through its lens. Through this ER bound, we show that clustering-based methods such as DeepCluster and SwAV maximize the MI. We also re-interpret the mechanisms of distillation-based approaches such as BYOL and DINO, showing that they explicitly maximize the reconstruction term and implicitly encourage a stable entropy, and we confirm this empirically. We show that replacing the objectives of common MVSSL methods with this ER bound achieves competitive performance, while making them stable when training with smaller batch sizes or smaller exponential moving average (EMA) coefficients.
