Approximated Oracle Filter Pruning for Destructive CNN Width Optimization

TL;DR

This paper introduces AOFP, a fast and effective method for CNN filter pruning that automatically finds optimal widths across layers, reducing computational costs while maintaining accuracy.

Contribution

AOFP presents a novel binary search-based pruning method that efficiently prunes multiple CNN layers simultaneously without heuristic guidance.

Findings

AOFP achieves significant model size reduction with minimal accuracy loss.

AOFP reduces pruning time compared to traditional oracle methods.

AOFP enables re-designing CNNs for higher accuracy and faster inference.

Abstract

It is not easy to design and run Convolutional Neural Networks (CNNs) due to: 1) finding the optimal number of filters (i.e., the width) at each layer is tricky, given an architecture; and 2) the computational intensity of CNNs impedes the deployment on computationally limited devices. Oracle Pruning is designed to remove the unimportant filters from a well-trained CNN, which estimates the filters' importance by ablating them in turn and evaluating the model, thus delivers high accuracy but suffers from intolerable time complexity, and requires a given resulting width but cannot automatically find it. To address these problems, we propose Approximated Oracle Filter Pruning (AOFP), which keeps searching for the least important filters in a binary search manner, makes pruning attempts by masking out filters randomly, accumulates the resulting errors, and finetunes the model via a multi-path framework. As AOFP enables simultaneous pruning on multiple layers, we can prune an existing very deep CNN with acceptable time cost, negligible accuracy drop, and no heuristic knowledge, or re-design a model which exerts higher accuracy and faster inference.