A Framework for Bilevel Optimization on Riemannian Manifolds

Abstract

Bilevel optimization has gained prominence in various applications. In this study, we introduce a framework for solving bilevel optimization problems, where the variables in both the lower and upper levels are constrained on Riemannian manifolds. We present several hypergradient estimation strategies on manifolds and analyze their estimation errors. Furthermore, we provide comprehensive convergence and complexity analyses for the proposed hypergradient descent algorithm on manifolds. We also extend our framework to encompass stochastic bilevel optimization and incorporate the use of general retraction. The efficacy of the proposed framework is demonstrated through several applications.

Cite

Text

Han et al. "A Framework for Bilevel Optimization on Riemannian Manifolds." Neural Information Processing Systems, 2024. doi:10.52202/079017-3299

Markdown

[Han et al. "A Framework for Bilevel Optimization on Riemannian Manifolds." Neural Information Processing Systems, 2024.](https://mlanthology.org/neurips/2024/han2024neurips-framework/) doi:10.52202/079017-3299

BibTeX

@inproceedings{han2024neurips-framework,
  title     = {{A Framework for Bilevel Optimization on Riemannian Manifolds}},
  author    = {Han, Andi and Mishra, Bamdev and Jawanpuria, Pratik and Takeda, Akiko},
  booktitle = {Neural Information Processing Systems},
  year      = {2024},
  doi       = {10.52202/079017-3299},
  url       = {https://mlanthology.org/neurips/2024/han2024neurips-framework/}
}