Dual Propagation: Accelerating Contrastive Hebbian Learning with Dyadic Neurons

TL;DR

This paper introduces dual propagation, a novel neuron model that enables faster and more energy-efficient contrastive Hebbian learning by reducing inference to a single layerwise process, matching back-propagation's accuracy.

Contribution

The authors propose a new energy-based compartmental neuron model called dual propagation, allowing single-phase inference and improved efficiency in contrastive Hebbian learning.

Findings

Performs comparably to back-propagation in accuracy.

Achieves faster inference with reduced computational cost.

Works effectively on datasets including Imagenet32x32.

Abstract

Activity difference based learning algorithms-such as contrastive Hebbian learning and equilibrium propagation-have been proposed as biologically plausible alternatives to error back-propagation. However, on traditional digital chips these algorithms suffer from having to solve a costly inference problem twice, making these approaches more than two orders of magnitude slower than back-propagation. In the analog realm equilibrium propagation may be promising for fast and energy efficient learning, but states still need to be inferred and stored twice. Inspired by lifted neural networks and compartmental neuron models we propose a simple energy based compartmental neuron model, termed dual propagation, in which each neuron is a dyad with two intrinsic states. At inference time these intrinsic states encode the error/activity duality through their difference and their mean respectively. The advantage of this method is that only a single inference phase is needed and that inference can be solved in layerwise closed-form. Experimentally we show on common computer vision datasets, including Imagenet32x32, that dual propagation performs equivalently to back-propagation both in terms of accuracy and runtime.