Pruning Randomly Initialized Neural Networks with Iterative Randomization
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
Cite
@article{arxiv.2106.09269,
title = {Pruning Randomly Initialized Neural Networks with Iterative Randomization},
author = {Daiki Chijiwa and Shin'ya Yamaguchi and Yasutoshi Ida and Kenji Umakoshi and Tomohiro Inoue},
journal= {arXiv preprint arXiv:2106.09269},
year = {2022}
}
Comments
35th Conference on Neural Information Processing Systems (NeurIPS 2021); Selected for a spotlight presentation