MineRL Diamond 2021 Competition: Overview, Results, and Lessons Learned

Anssi Kanervisto, Stephanie Milani, Karolis Ramanauskas, Nicholay Topin, Zichuan Lin, Junyou Li, Jianing Shi, Deheng Ye, Qiang Fu, Wei Yang, Weijun Hong, Zhongyue Huang, Haicheng Chen, Guangjun Zeng, Yue Lin, Vincent Micheli, Eloi Alonso, François Fleuret, Alexander Nikulin, Yury Belousov, Oleg Svidchenko, Aleksei Shpilman
Proceedings of the NeurIPS 2021 Competitions and Demonstrations Track, PMLR 176:13-28, 2022.

Abstract

Reinforcement learning competitions advance the field by providing appropriate scope and support to develop solutions toward a specific problem. To promote the development of more broadly applicable methods, organizers need to enforce the use of general techniques, the use of sample-efficient methods, and the reproducibility of the results. While beneficial for the research community, these restrictions come at a cost—increased difficulty. If the barrier for entry is too high, many potential participants are demoralized. With this in mind, we hosted the third edition of the MineRL ObtainDiamond competition, MineRL Diamond 2021, with a separate track in which we permitted any solution to promote the participation of newcomers. With this track and more extensive tutorials and support, we saw an increased number of submissions. The participants of this easier track were able to obtain a diamond, and the participants of the harder track progressed the generalizable solutions in the same task.

Cite this Paper

BibTeX
@InProceedings{pmlr-v176-kanervisto22a, title = {MineRL Diamond 2021 Competition: Overview, Results, and Lessons Learned}, author = {Kanervisto, Anssi and Milani, Stephanie and Ramanauskas, Karolis and Topin, Nicholay and Lin, Zichuan and Li, Junyou and Shi, Jianing and Ye, Deheng and Fu, Qiang and Yang, Wei and Hong, Weijun and Huang, Zhongyue and Chen, Haicheng and Zeng, Guangjun and Lin, Yue and Micheli, Vincent and Alonso, Eloi and Fleuret, Fran\c{c}ois and Nikulin, Alexander and Belousov, Yury and Svidchenko, Oleg and Shpilman, Aleksei}, booktitle = {Proceedings of the NeurIPS 2021 Competitions and Demonstrations Track}, pages = {13--28}, year = {2022}, editor = {Kiela, Douwe and Ciccone, Marco and Caputo, Barbara}, volume = {176}, series = {Proceedings of Machine Learning Research}, month = {06--14 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v176/kanervisto22a/kanervisto22a.pdf}, url = {https://proceedings.mlr.press/v176/kanervisto22a.html}, abstract = {Reinforcement learning competitions advance the field by providing appropriate scope and support to develop solutions toward a specific problem. To promote the development of more broadly applicable methods, organizers need to enforce the use of general techniques, the use of sample-efficient methods, and the reproducibility of the results. While beneficial for the research community, these restrictions come at a cost—increased difficulty. If the barrier for entry is too high, many potential participants are demoralized. With this in mind, we hosted the third edition of the MineRL ObtainDiamond competition, MineRL Diamond 2021, with a separate track in which we permitted any solution to promote the participation of newcomers. With this track and more extensive tutorials and support, we saw an increased number of submissions. The participants of this easier track were able to obtain a diamond, and the participants of the harder track progressed the generalizable solutions in the same task.} }
Endnote
%0 Conference Paper %T MineRL Diamond 2021 Competition: Overview, Results, and Lessons Learned %A Anssi Kanervisto %A Stephanie Milani %A Karolis Ramanauskas %A Nicholay Topin %A Zichuan Lin %A Junyou Li %A Jianing Shi %A Deheng Ye %A Qiang Fu %A Wei Yang %A Weijun Hong %A Zhongyue Huang %A Haicheng Chen %A Guangjun Zeng %A Yue Lin %A Vincent Micheli %A Eloi Alonso %A François Fleuret %A Alexander Nikulin %A Yury Belousov %A Oleg Svidchenko %A Aleksei Shpilman %B Proceedings of the NeurIPS 2021 Competitions and Demonstrations Track %C Proceedings of Machine Learning Research %D 2022 %E Douwe Kiela %E Marco Ciccone %E Barbara Caputo %F pmlr-v176-kanervisto22a %I PMLR %P 13--28 %U https://proceedings.mlr.press/v176/kanervisto22a.html %V 176 %X Reinforcement learning competitions advance the field by providing appropriate scope and support to develop solutions toward a specific problem. To promote the development of more broadly applicable methods, organizers need to enforce the use of general techniques, the use of sample-efficient methods, and the reproducibility of the results. While beneficial for the research community, these restrictions come at a cost—increased difficulty. If the barrier for entry is too high, many potential participants are demoralized. With this in mind, we hosted the third edition of the MineRL ObtainDiamond competition, MineRL Diamond 2021, with a separate track in which we permitted any solution to promote the participation of newcomers. With this track and more extensive tutorials and support, we saw an increased number of submissions. The participants of this easier track were able to obtain a diamond, and the participants of the harder track progressed the generalizable solutions in the same task.
APA
Kanervisto, A., Milani, S., Ramanauskas, K., Topin, N., Lin, Z., Li, J., Shi, J., Ye, D., Fu, Q., Yang, W., Hong, W., Huang, Z., Chen, H., Zeng, G., Lin, Y., Micheli, V., Alonso, E., Fleuret, F., Nikulin, A., Belousov, Y., Svidchenko, O. & Shpilman, A.. (2022). MineRL Diamond 2021 Competition: Overview, Results, and Lessons Learned. Proceedings of the NeurIPS 2021 Competitions and Demonstrations Track, in Proceedings of Machine Learning Research 176:13-28 Available from https://proceedings.mlr.press/v176/kanervisto22a.html.

Related Material