Gated Linear Networks

TL;DR

Gated Linear Networks (GLNs) are a new type of neural architecture that enables rapid online learning without backpropagation, offering universal learning capabilities and resilience to catastrophic forgetting, suitable for real-time applications.

Contribution

This paper introduces Gated Linear Networks, a novel neural architecture with local credit assignment and online convex optimization, distinct from traditional backpropagation-based models.

Findings

GLNs achieve universal learning in the limit as network size increases.

GLNs demonstrate strong resilience to catastrophic forgetting.

GLNs perform comparably to dropout and Elastic Weight Consolidation methods on benchmarks.

Abstract

This paper presents a new family of backpropagation-free neural architectures, Gated Linear Networks (GLNs). What distinguishes GLNs from contemporary neural networks is the distributed and local nature of their credit assignment mechanism; each neuron directly predicts the target, forgoing the ability to learn feature representations in favor of rapid online learning. Individual neurons can model nonlinear functions via the use of data-dependent gating in conjunction with online convex optimization. We show that this architecture gives rise to universal learning capabilities in the limit, with effective model capacity increasing as a function of network size in a manner comparable with deep ReLU networks. Furthermore, we demonstrate that the GLN learning mechanism possesses extraordinary resilience to catastrophic forgetting, performing comparably to a MLP with dropout and Elastic Weight Consolidation on standard benchmarks. These desirable theoretical and empirical properties position GLNs as a complementary technique to contemporary offline deep learning methods.