arXiv preprint ·

Two-State Dynamical Synapses as a Physical Substrate for Equilibrium Propagation

Dongshu Liu and Julie Grollier

Published
Source
arXiv preprint, posted before peer review
Original
Read it at arXiv
Topics
Emerging TechnologiesApplied Physics

Abstract

Equilibrium Propagation (EqProp) updates each synapse by comparing a local term at the free and nudged equilibria. Because these two values occur at different times, a physical implementation requires temporary local memory. We show that this memory can emerge from a two-timescale synaptic model in which neural and volatile synaptic dynamics are jointly derived from a single energy function. A volatile state $w$ relaxes toward the product of the pre- and postsynaptic activities, while a non-volatile state $M$ stores the learned parameter. At equilibrium, $w$ equals the local term entering the EqProp update. Using its net change between phases implements phase-based EqProp, while transferring its successive changes during nudging yields continual EqProp, in which the temporal comparison is carried by the synaptic dynamics themselves. On MNIST with a single-hidden-layer network, the phase-based model reaches $2.01 \pm 0.09\%$ test error versus $2.00 \pm 0.13\%$ for standard EqProp, and the continual model reaches $2.24 \pm 0.11\%$ versus $2.28 \pm 0.16\%$ for continual EqProp. The framework further specifies the dynamical requirements that a volatile--non-volatile synaptic device must satisfy to implement this learning mechanism.

The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.

Read it on ReadSearch

Each opens after you create an account, and brings you back to this paper.

Already a reader? Sign in

All new papers in Emerging Technologies

Get the few papers worth your time in Emerging Technologies every morning.

ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.

Papers belong to their authors and publishers; every page here links to the original. All research topics