Open Problem: Black-Box Reductions and Adaptive Gradient Methods for Nonconvex Optimization

Abstract

We describe an open problem: reduce offline nonconvex stochastic optimization to regret minimization in online convex optimization. The conjectured reduction aims to make progress on explaining the success of adaptive gradient methods for deep learning. A prize of 500 dollars is offered to the winner.

Cite

Text

Chen and Hazan. "Open Problem: Black-Box Reductions and Adaptive Gradient Methods for Nonconvex Optimization." Conference on Learning Theory, 2024.

Markdown

[Chen and Hazan. "Open Problem: Black-Box Reductions and Adaptive Gradient Methods for Nonconvex Optimization." Conference on Learning Theory, 2024.](https://mlanthology.org/colt/2024/chen2024colt-open/)

BibTeX

@inproceedings{chen2024colt-open,
  title     = {{Open Problem: Black-Box Reductions and Adaptive Gradient Methods for Nonconvex Optimization}},
  author    = {Chen, Xinyi and Hazan, Elad},
  booktitle = {Conference on Learning Theory},
  year      = {2024},
  pages     = {5317-5324},
  volume    = {247},
  url       = {https://mlanthology.org/colt/2024/chen2024colt-open/}
}