CaltechAUTHORS
  A Caltech Library Service

Learning to Search via Retrospective Imitation

Song, Jialin and Lanka, Ravi and Zhao, Albert and Yue, Yisong and Ono, Masahiro (2018) Learning to Search via Retrospective Imitation. . (Submitted) http://resolver.caltech.edu/CaltechAUTHORS:20190205-111204454

[img] PDF - Submitted Version
See Usage Policy.

1733Kb

Use this Persistent URL to link to this item: http://resolver.caltech.edu/CaltechAUTHORS:20190205-111204454

Abstract

We study the problem of learning a good search policy from demonstrations for combinatorial search spaces. We propose retrospective imitation learning, which, after initial training by an expert, improves itself by learning from its own retrospective solutions. That is, when the policy eventually reaches a feasible solution in a search tree after making mistakes and backtracks, it retrospectively constructs an improved search trace to the solution by removing backtracks, which is then used to further train the policy. A key feature of our approach is that it can iteratively scale up, or transfer, to larger problem sizes than the initial expert demonstrations, thus dramatically expanding its applicability beyond that of conventional imitation learning. We showcase the effectiveness of our approach on two tasks: synthetic maze solving, and integer program based risk-aware path planning.


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/1804.00846arXivDiscussion Paper
Record Number:CaltechAUTHORS:20190205-111204454
Persistent URL:http://resolver.caltech.edu/CaltechAUTHORS:20190205-111204454
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:92668
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:05 Feb 2019 19:14
Last Modified:05 Feb 2019 19:14

Repository Staff Only: item control page