Multi-Objective Population Based Training

Abstract

Population Based Training (PBT) is an efficient hyperparameter optimization algorithm. PBT is a single-objective algorithm, but many real-world hyperparameter optimization problems involve two or more conflicting objectives. In this work, we therefore introduce a multi-objective version of PBT, MO-PBT. Our experiments on diverse multi-objective hyperparameter optimization problems (Precision/Recall, Accuracy/Fairness, Accuracy/Adversarial Robustness) show that MO-PBT outperforms random search, single-objective PBT, and the state-of-the-art multi-objective hyperparameter optimization algorithm MO-ASHA.

Cite

Text

Dushatskiy et al. "Multi-Objective Population Based Training." International Conference on Machine Learning, 2023.

Markdown

[Dushatskiy et al. "Multi-Objective Population Based Training." International Conference on Machine Learning, 2023.](https://mlanthology.org/icml/2023/dushatskiy2023icml-multiobjective/)

BibTeX

@inproceedings{dushatskiy2023icml-multiobjective,
  title     = {{Multi-Objective Population Based Training}},
  author    = {Dushatskiy, Arkadiy and Chebykin, Alexander and Alderliesten, Tanja and Bosman, Peter},
  booktitle = {International Conference on Machine Learning},
  year      = {2023},
  pages     = {8969-8989},
  volume    = {202},
  url       = {https://mlanthology.org/icml/2023/dushatskiy2023icml-multiobjective/}
}