EigenDamage: Structured Pruning in the Kronecker-Factored Eigenbasis

TL;DR

This paper introduces EigenDamage, a structured pruning method using the Kronecker-factored eigenbasis to efficiently reduce neural network size and computation while maintaining accuracy.

Contribution

It proposes a novel KFE-based reparameterization and Hessian-based pruning approach that improves pruning accuracy and speed compared to existing methods.

Findings

Achieves 10x model size reduction with negligible accuracy loss.

Attains 8x FLOPs reduction on wide ResNet32.

More effective on challenging datasets and networks.

Abstract

Reducing the test time resource requirements of a neural network while preserving test accuracy is crucial for running inference on resource-constrained devices. To achieve this goal, we introduce a novel network reparameterization based on the Kronecker-factored eigenbasis (KFE), and then apply Hessian-based structured pruning methods in this basis. As opposed to existing Hessian-based pruning algorithms which do pruning in parameter coordinates, our method works in the KFE where different weights are approximately independent, enabling accurate pruning and fast computation. We demonstrate empirically the effectiveness of the proposed method through extensive experiments. In particular, we highlight that the improvements are especially significant for more challenging datasets and networks. With negligible loss of accuracy, an iterative-pruning version gives a 10$\times$ reduction in model size and a 8$\times$ reduction in FLOPs on wide ResNet32.