Group-Wise Oracle-Efficient Algorithms for Online Multi-Group Learning

Abstract

We study the problem of online multi-group learning, a learning model in which an online learner must simultaneously achieve small prediction regret on a large collection of (possibly overlapping) subsequences corresponding to a family of groups. Groups are subsets of the context space, and in fairness applications, they may correspond to subpopulations defined by expressive functions of demographic attributes. In this paper, we design such oracle-efficient algorithms with sublinear regret under a variety of settings, including: (i) the i.i.d. setting, (ii) the adversarial setting with smoothed context distributions, and (iii) the adversarial transductive setting.

PDF NeurIPS OpenReview Semantic Scholar

Cite

Text

Deng et al. "Group-Wise Oracle-Efficient Algorithms for Online Multi-Group Learning." Neural Information Processing Systems, 2024. doi:10.52202/079017-1247

Markdown

[Deng et al. "Group-Wise Oracle-Efficient Algorithms for Online Multi-Group Learning." Neural Information Processing Systems, 2024.](https://mlanthology.org/neurips/2024/deng2024neurips-groupwise/) doi:10.52202/079017-1247

BibTeX

@inproceedings{deng2024neurips-groupwise,
  title     = {{Group-Wise Oracle-Efficient Algorithms for Online Multi-Group Learning}},
  author    = {Deng, Samuel and Hsu, Daniel and Liu, Jingwen},
  booktitle = {Neural Information Processing Systems},
  year      = {2024},
  doi       = {10.52202/079017-1247},
  url       = {https://mlanthology.org/neurips/2024/deng2024neurips-groupwise/}
}