LinDeps: A Fine-tuning Free Post-Pruning Method to Remove Layer-Wise Linear Dependencies with Guaranteed Performance Preservation

TL;DR

LinDeps is a post-pruning method that efficiently removes redundant filters in CNNs by analyzing linear dependencies, preserving performance without fine-tuning, and enhancing existing pruning techniques.

Contribution

It introduces LinDeps, a novel post-pruning approach using linear dependency analysis with QR decomposition, applicable on top of any pruning method to improve compression and speed.

Findings

Achieves state-of-the-art pruning performance on CIFAR-10 and ImageNet.

Improves compression rates while maintaining accuracy.

Provides significant speedups in low-resource scenarios.

Abstract

Convolutional Neural Networks (CNN) are widely used in many computer vision tasks. Yet, their increasing size and complexity pose significant challenges for efficient deployment on resource-constrained platforms. Hence, network pruning has emerged as an effective way of reducing the size and computational requirements of neural networks by removing redundant or unimportant parameters. However, a fundamental challenge with pruning consists in optimally removing redundancies without degrading performance. Most existing pruning techniques overlook structural dependencies across feature maps within a layer, resulting in suboptimal pruning decisions. In this work, we introduce LinDeps, a novel post-pruning method, i.e., a pruning method that can be applied on top of any pruning technique, which systematically identifies and removes redundant filters via linear dependency analysis. Particularly, LinDeps applies pivoted QR decomposition to feature maps to detect and prune linearly dependent filters. Then, a novel signal recovery mechanism adjusts the next layer's kernels to preserve compatibility and performance without requiring any fine-tuning. Our experiments on CIFAR-10 and ImageNet with VGG and ResNet backbones demonstrate that LinDeps improves compression rates of existing pruning techniques while preserving performances, leading to a new state of the art in CNN pruning. We also benchmark LinDeps in low-resource setups where no retraining can be performed, which shows significant pruning improvements and inference speedups over a state-of-the-art method. LinDeps therefore constitutes an essential add-on for any current or future pruning technique.