Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation

Shani Gamrian, Yoav Goldberg
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:2063-2072, 2019.

Abstract

Despite the remarkable success of Deep RL in learning control policies from raw pixels, the resulting models do not generalize. We demonstrate that a trained agent fails completely when facing small visual changes, and that fine-tuning—the common transfer learning paradigm—fails to adapt to these changes, to the extent that it is faster to re-train the model from scratch. We show that by separating the visual transfer task from the control policy we achieve substantially better sample efficiency and transfer behavior, allowing an agent trained on the source task to transfer well to the target tasks. The visual mapping from the target to the source domain is performed using unaligned GANs, resulting in a control policy that can be further improved using imitation learning from imperfect demonstrations. We demonstrate the approach on synthetic visual variants of the Breakout game, as well as on transfer between subsequent levels of Road Fighter, a Nintendo car-driving game. A visualization of our approach can be seen in \url{https://youtu.be/4mnkzYyXMn4} and \url{https://youtu.be/KCGTrQi6Ogo}.

Cite this Paper


BibTeX
@InProceedings{pmlr-v97-gamrian19a, title = {Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation}, author = {Gamrian, Shani and Goldberg, Yoav}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {2063--2072}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/gamrian19a/gamrian19a.pdf}, url = {https://proceedings.mlr.press/v97/gamrian19a.html}, abstract = {Despite the remarkable success of Deep RL in learning control policies from raw pixels, the resulting models do not generalize. We demonstrate that a trained agent fails completely when facing small visual changes, and that fine-tuning—the common transfer learning paradigm—fails to adapt to these changes, to the extent that it is faster to re-train the model from scratch. We show that by separating the visual transfer task from the control policy we achieve substantially better sample efficiency and transfer behavior, allowing an agent trained on the source task to transfer well to the target tasks. The visual mapping from the target to the source domain is performed using unaligned GANs, resulting in a control policy that can be further improved using imitation learning from imperfect demonstrations. We demonstrate the approach on synthetic visual variants of the Breakout game, as well as on transfer between subsequent levels of Road Fighter, a Nintendo car-driving game. A visualization of our approach can be seen in \url{https://youtu.be/4mnkzYyXMn4} and \url{https://youtu.be/KCGTrQi6Ogo}.} }
Endnote
%0 Conference Paper %T Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation %A Shani Gamrian %A Yoav Goldberg %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-gamrian19a %I PMLR %P 2063--2072 %U https://proceedings.mlr.press/v97/gamrian19a.html %V 97 %X Despite the remarkable success of Deep RL in learning control policies from raw pixels, the resulting models do not generalize. We demonstrate that a trained agent fails completely when facing small visual changes, and that fine-tuning—the common transfer learning paradigm—fails to adapt to these changes, to the extent that it is faster to re-train the model from scratch. We show that by separating the visual transfer task from the control policy we achieve substantially better sample efficiency and transfer behavior, allowing an agent trained on the source task to transfer well to the target tasks. The visual mapping from the target to the source domain is performed using unaligned GANs, resulting in a control policy that can be further improved using imitation learning from imperfect demonstrations. We demonstrate the approach on synthetic visual variants of the Breakout game, as well as on transfer between subsequent levels of Road Fighter, a Nintendo car-driving game. A visualization of our approach can be seen in \url{https://youtu.be/4mnkzYyXMn4} and \url{https://youtu.be/KCGTrQi6Ogo}.
APA
Gamrian, S. & Goldberg, Y.. (2019). Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:2063-2072 Available from https://proceedings.mlr.press/v97/gamrian19a.html.

Related Material