Channel-Directed Gradients for Optimization of Convolutional Neural Networks

TL;DR

This paper proposes a simple, efficient method to improve the generalization of convolutional neural networks by modifying gradients along output-channel directions, applicable with any optimizer.

Contribution

It introduces a novel gradient modification technique based on output-channel directed metrics that enhances CNN training without significant computational overhead.

Findings

Improved generalization error across multiple benchmarks.

Gradient smoothing along output channels boosts performance.

Method compatible with existing optimizers and architectures.

Abstract

We introduce optimization methods for convolutional neural networks that can be used to improve existing gradient-based optimization in terms of generalization error. The method requires only simple processing of existing stochastic gradients, can be used in conjunction with any optimizer, and has only a linear overhead (in the number of parameters) compared to computation of the stochastic gradient. The method works by computing the gradient of the loss function with respect to output-channel directed re-weighted L2 or Sobolev metrics, which has the effect of smoothing components of the gradient across a certain direction of the parameter tensor. We show that defining the gradients along the output channel direction leads to a performance boost, while other directions can be detrimental. We present the continuum theory of such gradients, its discretization, and application to deep networks. Experiments on benchmark datasets, several networks and baseline optimizers show that optimizers can be improved in generalization error by simply computing the stochastic gradient with respect to output-channel directed metrics.