Bag of Baselines for Multi-Objective Joint Neural Architecture Search and Hyperparameter Optimization
ICML Workshop on Automated Machine Learning (AutoML), 2021
Abstract: While both neural architecture search (NAS) and hyperparameter optimization (HPO) have been studied extensively in recent years, NAS methods typically assume fixed hyperparameters and vice versa. Furthermore, NAS has recently often been framed as a multi-objective optimization problem, in order to take, e.g., resource requirements into account. In this paper, we propose a set of methods that extend current approaches to jointly optimize neural architectures and hyperparameters with respect to multiple objectives. We hope that these methods will serve as simple baselines for future research on multi-objective joint NAS + HPO.
Paper
Downloads
Images and movies
![thumbnail.jpg [38KB]](/Publications/2021/Sch21/thumbnail.jpg "thumbnail.jpg [38KB]")
BibTex reference
@InProceedings{Sch21,
author = "S. Izquierdo and J. Guerrero-Viu and S. Hauns and G. Miotto and S. Schrodi and T. Elsken and D. Deng and M. Lindauer and F. Hutter",
title = "Bag of Baselines for Multi-Objective Joint Neural Architecture Search and Hyperparameter Optimization",
booktitle = "ICML Workshop on Automated Machine Learning (AutoML)",
month = " ",
year = "2021",
url = "http://lmb.informatik.uni-freiburg.de/Publications/2021/Sch21"
}