Playing Minecraft with Behavioural Cloning

Anssi Kanervisto, Janne Karttunen, Ville Hautamäki
Proceedings of the NeurIPS 2019 Competition and Demonstration Track, PMLR 123:56-66, 2020.

Abstract

MineRL 2019 competition challenged participants to train sample-efficient agents to play Minecraft, by using a dataset of human gameplay and a limit number of steps the environment. We approached this task with behavioural cloning by predicting what actions human players would take, and reached fifth place in the final ranking. Despite being a simple algorithm, we observed the performance of such an approach can vary significantly, based on when the training is stopped. In this paper, we detail our submission to the competition, run further experiments to study how performance varied over training and study how different engineering decisions affected these results.

Cite this Paper

BibTeX
@InProceedings{pmlr-v123-kanervisto20a, title = {Playing Minecraft with Behavioural Cloning}, author = {Kanervisto, Anssi and Karttunen, Janne and Hautam\"aki, Ville}, booktitle = {Proceedings of the NeurIPS 2019 Competition and Demonstration Track}, pages = {56--66}, year = {2020}, editor = {Escalante, Hugo Jair and Hadsell, Raia}, volume = {123}, series = {Proceedings of Machine Learning Research}, month = {08--14 Dec}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v123/kanervisto20a/kanervisto20a.pdf}, url = {https://proceedings.mlr.press/v123/kanervisto20a.html}, abstract = {MineRL 2019 competition challenged participants to train sample-efficient agents to play Minecraft, by using a dataset of human gameplay and a limit number of steps the environment. We approached this task with behavioural cloning by predicting what actions human players would take, and reached fifth place in the final ranking. Despite being a simple algorithm, we observed the performance of such an approach can vary significantly, based on when the training is stopped. In this paper, we detail our submission to the competition, run further experiments to study how performance varied over training and study how different engineering decisions affected these results.} }
Endnote
%0 Conference Paper %T Playing Minecraft with Behavioural Cloning %A Anssi Kanervisto %A Janne Karttunen %A Ville Hautamäki %B Proceedings of the NeurIPS 2019 Competition and Demonstration Track %C Proceedings of Machine Learning Research %D 2020 %E Hugo Jair Escalante %E Raia Hadsell %F pmlr-v123-kanervisto20a %I PMLR %P 56--66 %U https://proceedings.mlr.press/v123/kanervisto20a.html %V 123 %X MineRL 2019 competition challenged participants to train sample-efficient agents to play Minecraft, by using a dataset of human gameplay and a limit number of steps the environment. We approached this task with behavioural cloning by predicting what actions human players would take, and reached fifth place in the final ranking. Despite being a simple algorithm, we observed the performance of such an approach can vary significantly, based on when the training is stopped. In this paper, we detail our submission to the competition, run further experiments to study how performance varied over training and study how different engineering decisions affected these results.
APA
Kanervisto, A., Karttunen, J. & Hautamäki, V.. (2020). Playing Minecraft with Behavioural Cloning. Proceedings of the NeurIPS 2019 Competition and Demonstration Track, in Proceedings of Machine Learning Research 123:56-66 Available from https://proceedings.mlr.press/v123/kanervisto20a.html.

Related Material