References & Citations
Computer Science > Artificial Intelligence
Title: Marginal Utility for Planning in Continuous or Large Discrete Action Spaces
(Submitted on 10 Jun 2020 (v1), last revised 17 Jun 2020 (this version, v2))
Abstract: Sample-based planning is a powerful family of algorithms for generating intelligent behavior from a model of the environment. Generating good candidate actions is critical to the success of sample-based planners, particularly in continuous or large action spaces. Typically, candidate action generation exhausts the action space, uses domain knowledge, or more recently, involves learning a stochastic policy to provide such search guidance. In this paper we explore explicitly learning a candidate action generator by optimizing a novel objective, marginal utility. The marginal utility of an action generator measures the increase in value of an action over previously generated actions. We validate our approach in both curling, a challenging stochastic domain with continuous state and action spaces, and a location game with a discrete but large action space. We show that a generator trained with the marginal utility objective outperforms hand-coded schemes built on substantial domain knowledge, trained stochastic policies, and other natural objectives for generating actions for sampled-based planners.
Submission history
From: Zaheen Ahmad [view email][v1] Wed, 10 Jun 2020 20:24:53 GMT (672kb,D)
[v2] Wed, 17 Jun 2020 17:00:39 GMT (673kb,D)
Link back to: arXiv, form interface, contact.