Neuron Pruning for Compressing Deep Networks using Maxout Architectures

TL;DR

This paper introduces a neuron pruning method using maxout units that significantly reduces deep neural network sizes while maintaining performance, and can be combined with weight pruning for even greater compression.

Contribution

It proposes a novel neuron pruning approach leveraging maxout architectures and local relevance measures, demonstrating substantial parameter reduction without performance loss.

Findings

Neuron pruning reduces network parameters by up to 74% and 61% on MNIST and LFW.

Combining neuron and weight pruning achieves up to 92% and 80% reduction.

The method maintains network accuracy despite significant size reduction.

Abstract

This paper presents an efficient and robust approach for reducing the size of deep neural networks by pruning entire neurons. It exploits maxout units for combining neurons into more complex convex functions and it makes use of a local relevance measurement that ranks neurons according to their activation on the training set for pruning them. Additionally, a parameter reduction comparison between neuron and weight pruning is shown. It will be empirically shown that the proposed neuron pruning reduces the number of parameters dramatically. The evaluation is performed on two tasks, the MNIST handwritten digit recognition and the LFW face verification, using a LeNet-5 and a VGG16 network architecture. The network size is reduced by up to $74\%$ and $61\%$, respectively, without affecting the network's performance. The main advantage of neuron pruning is its direct influence on the size of the network architecture. Furthermore, it will be shown that neuron pruning can be combined with subsequent weight pruning, reducing the size of the LeNet-5 and VGG16 up to $92\%$ and $80\%$ respectively.