We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.AI

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Artificial Intelligence

Title: Reinforcement Learning for Classical Planning: Viewing Heuristics as Dense Reward Generators

Abstract: Recent advances in reinforcement learning (RL) have led to a growing interest in applying RL to classical planning domains or applying classical planning methods to some complex RL domains. However, the long-horizon goal-based problems found in classical planning lead to sparse rewards for RL, making direct application inefficient. In this paper, we propose to leverage domain-independent heuristic functions commonly used in the classical planning literature to improve the sample efficiency of RL. These classical heuristics act as dense reward generators to alleviate the sparse-rewards issue and enable our RL agent to learn domain-specific value functions as residuals on these heuristics, making learning easier. Correct application of this technique requires consolidating the discounted metric used in RL and the non-discounted metric used in heuristics. We implement the value functions using Neural Logic Machines, a neural network architecture designed for grounded first-order logic inputs. We demonstrate on several classical planning domains that using classical heuristics for RL allows for good sample efficiency compared to sparse-reward RL. We further show that our learned value functions generalize to novel problem instances in the same domain.
Comments: Equal contributions by the first two authors. This manuscript is a camera-ready version accepted in ICAPS-2022. It is significantly updated from past versions (e.g., in the ICAPS PRL (Planning and RL) workshop) with additional experiments comparing existing work (STRIPS-HGN (Shen, Trevizan, and Thiebaux 2020) and GBFS-GNN (Rivlin, Hazan, and Karpas 2019))
Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Cite as: arXiv:2109.14830 [cs.AI]
  (or arXiv:2109.14830v2 [cs.AI] for this version)

Submission history

From: Masataro Asai [view email]
[v1] Thu, 30 Sep 2021 03:36:01 GMT (9246kb,D)
[v2] Mon, 7 Mar 2022 18:51:01 GMT (5184kb,D)

Link back to: arXiv, form interface, contact.