How Well Do Sparse Imagenet Models Transfer?

TL;DR

This paper investigates how sparsified ImageNet-trained CNN models transfer to downstream tasks, finding that sparse models can match or outperform dense models in transfer accuracy while offering speedups.

Contribution

It provides an extensive analysis of transfer performance of various state-of-the-art pruning methods on CNNs trained on ImageNet across multiple tasks.

Findings

Sparse models match or outperform dense models in transfer accuracy.

Pruned models can achieve significant inference and training speedups.

Different pruning methods exhibit distinct transfer behaviors.

Abstract

Transfer learning is a classic paradigm by which models pretrained on large "upstream" datasets are adapted to yield good results on "downstream" specialized datasets. Generally, more accurate models on the "upstream" dataset tend to provide better transfer accuracy "downstream". In this work, we perform an in-depth investigation of this phenomenon in the context of convolutional neural networks (CNNs) trained on the ImageNet dataset, which have been pruned - that is, compressed by sparsifying their connections. We consider transfer using unstructured pruned models obtained by applying several state-of-the-art pruning methods, including magnitude-based, second-order, re-growth, lottery-ticket, and regularization approaches, in the context of twelve standard transfer tasks. In a nutshell, our study shows that sparse models can match or even outperform the transfer performance of dense models, even at high sparsities, and, while doing so, can lead to significant inference and even training speedups. At the same time, we observe and analyze significant differences in the behaviour of different pruning methods.