Building Efficient ConvNets using Redundant Feature Pruning

TL;DR

This paper introduces a method for pruning redundant features in convolutional neural networks by analyzing feature similarity, significantly reducing inference costs while maintaining competitive accuracy.

Contribution

It proposes a novel feature pruning technique based on cosine distance to eliminate redundant filters, improving efficiency of deep CNNs.

Findings

Reduced inference costs by up to 40% on VGG-16

Achieved 27% reduction on ResNet-56

Maintained competitive performance after pruning

Abstract

This paper presents an efficient technique to prune deep and/or wide convolutional neural network models by eliminating redundant features (or filters). Previous studies have shown that over-sized deep neural network models tend to produce a lot of redundant features that are either shifted version of one another or are very similar and show little or no variations; thus resulting in filtering redundancy. We propose to prune these redundant features along with their connecting feature maps according to their differentiation and based on their relative cosine distances in the feature space, thus yielding smaller network size with reduced inference costs and competitive performance. We empirically show on select models and CIFAR-10 dataset that inference costs can be reduced by 40% for VGG-16, 27% for ResNet-56, and 39% for ResNet-110.