Problem Decomposition for Behavioural Cloning
Abstract
In behavioural cloning of the human operator’s skill, a controller is usually induced directly as a classifier from system’s states into actions. Experience shows that this often results in brittle controllers. In this paper we explore a decomposition of the cloning problem into two learning problems: the learning of operator’s control trajectories and the learning of the system’s dynamics separately. We analyse advantages of such indirect controllers . We give characterization of the learner’s error that is plausible explanation of why this decomposition approach has empirically proved to be usually superior to direct cloning.
Cite
Text
Suc and Bratko. "Problem Decomposition for Behavioural Cloning." European Conference on Machine Learning, 2000. doi:10.1007/3-540-45164-1_39Markdown
[Suc and Bratko. "Problem Decomposition for Behavioural Cloning." European Conference on Machine Learning, 2000.](https://mlanthology.org/ecmlpkdd/2000/suc2000ecml-problem/) doi:10.1007/3-540-45164-1_39BibTeX
@inproceedings{suc2000ecml-problem,
title = {{Problem Decomposition for Behavioural Cloning}},
author = {Suc, Dorian and Bratko, Ivan},
booktitle = {European Conference on Machine Learning},
year = {2000},
pages = {382-391},
doi = {10.1007/3-540-45164-1_39},
url = {https://mlanthology.org/ecmlpkdd/2000/suc2000ecml-problem/}
}