Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons

TL;DR

This paper introduces Latent Equilibrium, a novel framework enabling rapid inference and learning in slow neural systems by leveraging phase-advancement, thus overcoming delays inherent in biological and physical networks.

Contribution

It proposes a biologically plausible model that approximates error backpropagation in deep cortical networks using continuous-time dynamics and local plasticity, avoiding phased learning phases.

Findings

Achieves competitive performance on benchmark datasets.

Demonstrates robustness to substrate imperfections.

Enables quasi-instantaneous inference regardless of network depth.

Abstract

The response time of physical computational elements is finite, and neurons are no exception. In hierarchical models of cortical networks each layer thus introduces a response lag. This inherent property of physical dynamical systems results in delayed processing of stimuli and causes a timing mismatch between network output and instructive signals, thus afflicting not only inference, but also learning. We introduce Latent Equilibrium, a new framework for inference and learning in networks of slow components which avoids these issues by harnessing the ability of biological neurons to phase-advance their output with respect to their membrane potential. This principle enables quasi-instantaneous inference independent of network depth and avoids the need for phased plasticity or computationally expensive network relaxation phases. We jointly derive disentangled neuron and synapse dynamics from a prospective energy function that depends on a network's generalized position and momentum. The resulting model can be interpreted as a biologically plausible approximation of error backpropagation in deep cortical networks with continuous-time, leaky neuronal dynamics and continuously active, local plasticity. We demonstrate successful learning of standard benchmark datasets, achieving competitive performance using both fully-connected and convolutional architectures, and show how our principle can be applied to detailed models of cortical microcircuitry. Furthermore, we study the robustness of our model to spatio-temporal substrate imperfections to demonstrate its feasibility for physical realization, be it in vivo or in silico.