Miniaturized Graph Convolutional Networks with Topologically Consistent Pruning

TL;DR

This paper introduces a novel magnitude pruning method for graph convolutional networks that guarantees topological consistency, leading to more effective and generalizable lightweight models especially under high pruning regimes.

Contribution

A new magnitude pruning approach with a reparametrization and bi-directional supervision that ensures topologically consistent subnetworks in graph neural networks.

Findings

Enhanced generalization under high pruning regimes

Maintained connectivity in pruned subnetworks

Improved performance on skeleton-based action recognition

Abstract

Magnitude pruning is one of the mainstream methods in lightweight architecture design whose goal is to extract subnetworks with the largest weight connections. This method is known to be successful, but under very high pruning regimes, it suffers from topological inconsistency which renders the extracted subnetworks disconnected, and this hinders their generalization ability. In this paper, we devise a novel magnitude pruning method that allows extracting subnetworks while guarantying their topological consistency. The latter ensures that only accessible and co-accessible -- impactful -- connections are kept in the resulting lightweight networks. Our solution is based on a novel reparametrization and two supervisory bi-directional networks which implement accessibility/co-accessibility and guarantee that only connected subnetworks will be selected during training. This solution allows enhancing generalization significantly, under very high pruning regimes, as corroborated through extensive experiments, involving graph convolutional networks, on the challenging task of skeleton-based action recognition.