Temporal Contrastive Learning through implicit non-equilibrium memory

TL;DR

This paper introduces Temporal Contrastive Learning, a local learning method using implicit non-equilibrium memory via integral feedback, enabling contrastive learning in physical and biological systems with energy considerations.

Contribution

It proposes a novel contrastive learning approach that employs implicit memory through integral feedback, broadening applicability to physical and biological systems.

Findings

Implicit non-equilibrium memory enables contrastive learning.

Non-equilibrium dissipation enhances learning quality.

Energy cost of contrastive learning is characterized.

Abstract

The backpropagation method has enabled transformative uses of neural networks. Alternatively, for energy-based models, local learning methods involving only nearby neurons offer benefits in terms of decentralized training, and allow for the possibility of learning in computationally-constrained substrates. One class of local learning methods constrasts the desired, clamped behavior with spontaneous, free behavior. However, directly contrasting free and clamped behaviors requires explicit memory. Here, we introduce `Temporal Contrastive Learning', an approach that uses integral feedback in each learning degree of freedom to provide a simple form of implicit non-equilibrium memory. During training, free and clamped behaviors are shown in a sawtooth-like protocol over time. When combined with integral feedback dynamics, these alternating temporal protocols generate an implicit memory necessary for comparing free and clamped behaviors, broadening the range of physical and biological systems capable of contrastive learning. Finally, we show that non-equilibrium dissipation improves learning quality and determine a Landauer-like energy cost of contrastive learning through physical dynamics.