Why Random Pruning Is All We Need to Start Sparse

(2023) Why Random Pruning Is All We Need to Start Sparse.

Text 2441_why_random_pruning_is_all_we_n.pdf Download (1MB)

Abstract

Random masks define surprisingly effective sparse neural network models, as has been shown empirically. The resulting sparse networks can often compete with dense architectures and state-of-the-art lottery ticket pruning algorithms, even though they do not rely on computationally expensive prune-train iterations and can be drawn initially without significant computational over-head. We offer a theoretical explanation of how random masks can approximate arbitrary target networks if they are wider by a logarithmic factor in the inverse sparsity 1/ log(1/sparsity). This overparameterization factor is necessary at least for 3-layer random networks, which elucidates the observed degrading performance of random networks at higher sparsity. At moderate to high sparsity levels, however, our results imply that sparser networks are contained within random source networks so that any dense-to-sparse training scheme can be turned into a computationally more efficient sparse to sparse one by constraining the search to a fixed random mask. We demonstrate the feasibility of this approach in experiments for different pruning methods and propose particularly effective choices of initial layer-wise sparsity ratios of the random source network. As a special case, we show theoretically and experimentally that random source networks also contain strong lottery tickets. Our code is available at https://github.com/RelationalML/sparse_to_sparse.

Item Type:	Conference or Workshop Item (A Paper) (Paper)
Divisions:	Rebekka Burkholz (RB)
Conference:	ICML International Conference on Machine Learning
Depositing User:	Rebekka Burkholz
Date Deposited:	06 Sep 2023 08:15
Last Modified:	06 Sep 2023 08:16
Primary Research Area:	NRA1: Trustworthy Information Processing
URI:	https://publications.cispa.saarland/id/eprint/4026

Actions

Actions (login required)

View Item