Associative Memory System via Threshold Linear Networks

TL;DR

This paper introduces a novel online auto-associative memory system using threshold-linear networks, enabling sequential memory formation with formal guarantees of robust retrieval from corrupted inputs.

Contribution

It presents a new memory system supporting sequential learning and provides formal guarantees for robust pattern retrieval, unlike previous models.

Findings

Successfully reconstructs patterns from corrupted inputs in simulations.

Supports sequential memory formation with formal guarantees.

Uses a threshold-linear network as latent space dynamics.

Abstract

Humans learn and form memories in stochastic environments. Auto-associative memory systems model these processes by storing patterns and later recovering them from corrupted versions. Here, memories are learned by associating each pattern with an attractor in a latent space. After learning, when (possibly corrupted) patterns are presented to the system, latent dynamics facilitate retrieval of the appropriate uncorrupted pattern. In this work, we propose a novel online auto-associative memory system. In contrast to existing works, our system supports sequential memory formation and provides formal guarantees of robust memory retrieval via region-of-attraction analysis. We use a threshold-linear network as latent space dynamics in combination with an encoder, decoder, and controller. We show in simulation that the memory system successfully reconstructs patterns from corrupted inputs.