DropNet: Reducing Neural Network Complexity via Iterative Pruning

TL;DR

DropNet is an iterative pruning method that significantly reduces neural network complexity by removing nodes and filters with minimal impact on accuracy, making models more efficient for edge deployment.

Contribution

This paper introduces DropNet, a novel iterative pruning technique that effectively reduces network size while maintaining performance, inspired by the Lottery Ticket Hypothesis.

Findings

Up to 90% of nodes/filters can be pruned without accuracy loss.

DropNet performs well across MLPs and CNNs on multiple datasets.

Pruned networks retain accuracy even after reinitialization.

Abstract

Modern deep neural networks require a significant amount of computing time and power to train and deploy, which limits their usage on edge devices. Inspired by the iterative weight pruning in the Lottery Ticket Hypothesis, we propose DropNet, an iterative pruning method which prunes nodes/filters to reduce network complexity. DropNet iteratively removes nodes/filters with the lowest average post-activation value across all training samples. Empirically, we show that DropNet is robust across diverse scenarios, including MLPs and CNNs using the MNIST, CIFAR-10 and Tiny ImageNet datasets. We show that up to 90% of the nodes/filters can be removed without any significant loss of accuracy. The final pruned network performs well even with reinitialization of the weights and biases. DropNet also has similar accuracy to an oracle which greedily removes nodes/filters one at a time to minimise training loss, highlighting its effectiveness.