Pruning Randomly Initialized Neural Networks with Iterative Randomization

TL;DR

This paper introduces IteRand, a novel framework for pruning randomly initialized neural networks through iterative randomization of weights, reducing parameter requirements while maintaining performance, supported by theoretical proofs and empirical results.

Contribution

The paper proposes a new iterative randomization framework for pruning neural networks, improving parameter efficiency over previous methods.

Findings

Theoretically proves effectiveness of randomization in reducing parameters.

Empirically demonstrates parameter efficiency on CIFAR-10 and ImageNet.

Achieves comparable performance with fewer parameters.

Abstract

Pruning the weights of randomly initialized neural networks plays an important role in the context of lottery ticket hypothesis. Ramanujan et al. (2020) empirically showed that only pruning the weights can achieve remarkable performance instead of optimizing the weight values. However, to achieve the same level of performance as the weight optimization, the pruning approach requires more parameters in the networks before pruning and thus more memory space. To overcome this parameter inefficiency, we introduce a novel framework to prune randomly initialized neural networks with iteratively randomizing weight values (IteRand). Theoretically, we prove an approximation theorem in our framework, which indicates that the randomizing operations are provably effective to reduce the required number of the parameters. We also empirically demonstrate the parameter efficiency in multiple experiments on CIFAR-10 and ImageNet. The code is available at: https://github.com/dchiji-ntt/iterand