English

Nystrom Method for Accurate and Scalable Implicit Differentiation

Machine Learning 2023-02-21 v1

Abstract

The essential difficulty of gradient-based bilevel optimization using implicit differentiation is to estimate the inverse Hessian vector product with respect to neural network parameters. This paper proposes to tackle this problem by the Nystrom method and the Woodbury matrix identity, exploiting the low-rankness of the Hessian. Compared to existing methods using iterative approximation, such as conjugate gradient and the Neumann series approximation, the proposed method avoids numerical instability and can be efficiently computed in matrix operations without iterations. As a result, the proposed method works stably in various tasks and is faster than iterative approximations. Throughout experiments including large-scale hyperparameter optimization and meta learning, we demonstrate that the Nystrom method consistently achieves comparable or even superior performance to other approaches. The source code is available from https://github.com/moskomule/hypergrad.

Keywords

Cite

@article{arxiv.2302.09726,
  title  = {Nystrom Method for Accurate and Scalable Implicit Differentiation},
  author = {Ryuichiro Hataya and Makoto Yamada},
  journal= {arXiv preprint arXiv:2302.09726},
  year   = {2023}
}

Comments

AISTATS 2023

R2 v1 2026-06-28T08:44:04.962Z