Sign-In to the Lottery: Reparameterizing Sparse Training

Advait Gadhikar, Tom Jacobs, Chao Zhou, Rebekka Burkholz

Keywords: pruning at initialization, sparse training, lottery ticket hypothesis, mirror flow, reparameterization, sign flips

TL;DR: Reparameterizing sparse training for improved PaI by enabling sign flips

Abstract: The performance gap between training sparse neural networks from scratch (PaI) and dense-to-sparse training presents a major roadblock for efficient deep learning. According to the Lottery Ticket Hypothesis, PaI hinges on finding a problem specific parameter initialization. As we show, to this end, determining correct parameter signs is sufficient. Yet, they remain elusive to PaI. To address this issue, we propose Sign-In, which employs a dynamic reparameterization that provably induces sign flips. Such sign flips are complementary to the ones that dense-to-sparse training can accomplish, rendering Sign-In as an orthogonal method. While our experiments and theory suggest performance improvements of PaI, they also carve out the main open challenge to close the gap between PaI and dense-to-sparse training.

Primary Area: Deep learning (e.g., architectures, generative models, optimization for deep networks, foundation models, LLMs)

Submission Number: 5689