CaltechAUTHORS
  A Caltech Library Service

Smooth Imitation Learning for Online Sequence Prediction

Le, Hoang M. and Kang, Andrew and Yue, Yisong and Carr, Peter (2016) Smooth Imitation Learning for Online Sequence Prediction. Proceedings of Machine Learning Research, 48 . pp. 680-688. ISSN 1938-7228. http://resolver.caltech.edu/CaltechAUTHORS:20160628-161544348

[img] PDF - Published Version
See Usage Policy.

942Kb
[img] PDF - Supplemental Material
See Usage Policy.

380Kb
[img] PDF - Submitted Version
See Usage Policy.

1103Kb

Use this Persistent URL to link to this item: http://resolver.caltech.edu/CaltechAUTHORS:20160628-161544348

Abstract

We study the problem of smooth imitation learning for online sequence prediction, where the goal is to train a policy that can smoothly imitate demonstrated behavior in a dynamic and continuous environment in response to online, sequential context input. Since the mapping from context to behavior is often complex, we take a learning reduction approach to reduce smooth imitation learning to a regression problem using complex function classes that are regularized to ensure smoothness. We present a learning meta-algorithm that achieves fast and stable convergence to a good policy. Our approach enjoys several attractive properties, including being fully deterministic, employing an adaptive learning rate that can provably yield larger policy improvements compared to previous approaches, and the ability to ensure stable convergence. Our empirical results demonstrate significant performance gains over previous approaches.


Item Type:Article
Related URLs:
URLURL TypeDescription
http://proceedings.mlr.press/v48/le16.htmlPublisherArticle
http://jmlr.org/proceedings/papers/v48/le16-supp.pdfPublisherSupplementary Material
https://arxiv.org/abs/1606.00968arXivDiscussion Paper
ORCID:
AuthorORCID
Yue, Yisong0000-0001-9127-1989
Additional Information:© 2016 by the author(s).
Record Number:CaltechAUTHORS:20160628-161544348
Persistent URL:http://resolver.caltech.edu/CaltechAUTHORS:20160628-161544348
Official Citation:Hoang Le, Andrew Kang, Yisong Yue, Peter Carr ; Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:680-688, 2016.
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:68723
Collection:CaltechAUTHORS
Deposited By: Joy Painter
Deposited On:28 Jun 2016 23:25
Last Modified:27 Mar 2019 16:42

Repository Staff Only: item control page