The code for reproducing the experiments described in
@article{adriaensen-arxiv22,
title={Automated Dynamic Algorithm Configuration},
author={Adriaensen, Steven and Biedenkapp, Andr{\'{e}} and Shala, Gresa and Awad, Noor and Eimer, Theresa and Lindauer, Marius and Hutter, Frank},
journal={arXiv:2205.13881 [cs.AI]},
year={2022},
}
Please cite this paper if you make use of this code (full pdf).
This repository contains three folders, each containing the code for the three case studies considered in the paper:
cmaes
: Step Size Adaptation in CMA-ESfd
: Heuristic Selection in FastDownwardsgd
: Learning Rate Control in Neural Network Training
To get started, clone the repository (git clone git@github.com:automl/2021_JAIR_DAC_experiments.git
) and cd
into the specific folder and see the respective README.md
's for instructions on how to use this code.