Deep Rewiring: Training very sparse deep networks

TL;DR

DEEP R is a novel algorithm that trains sparse neural networks by dynamically rewiring connections during supervised learning, maintaining strict sparsity bounds while achieving near-regular performance on benchmark tasks.

Contribution

It introduces a theoretically grounded method for training sparse networks through automatic rewiring, enabling efficient deep learning on hardware with connectivity constraints.

Findings

Effective training of very sparse networks with minor performance loss

Applicable to both feedforward and recurrent architectures

Based on stochastic sampling from a posterior distribution

Abstract

Neuromorphic hardware tends to pose limits on the connectivity of deep networks that one can run on them. But also generic hardware and software implementations of deep learning run more efficiently for sparse networks. Several methods exist for pruning connections of a neural network after it was trained without connectivity constraints. We present an algorithm, DEEP R, that enables us to train directly a sparsely connected neural network. DEEP R automatically rewires the network during supervised training so that connections are there where they are most needed for the task, while its total number is all the time strictly bounded. We demonstrate that DEEP R can be used to train very sparse feedforward and recurrent neural networks on standard benchmark tasks with just a minor loss in performance. DEEP R is based on a rigorous theoretical foundation that views rewiring as stochastic sampling of network configurations from a posterior.