Action Schema Networks: Generalised Policies with Deep Learning

Abstract

In this paper, we introduce the Action Schema Network (ASNet): a neural network architecture for learning generalised policies for probabilistic planning problems. By mimicking the relational structure of planning problems, ASNets are able to adopt a weight-sharing scheme which allows the network to be applied to any problem from a given planning domain. This allows the cost of training the network to be amortised over all problems in that domain. Further, we propose a training method which balances exploration and supervised training on small problems to produce a policy which remains robust when evaluated on larger problems. In experiments, we show that ASNet's learning capability allows it to significantly outperform traditional non-learning planners in several challenging domains.

Accepted to AAAI 2018

Cited by in corpus (10)

• Generalized Planning With Deep Reinforcement Learning
• Learning Domain-Independent Planning Heuristics with Hypergraph Networks
• Qualitative Numeric Planning: Reductions and Complexity
• Learning General Policies from Small Examples Without Supervision
• Symbolic Network: Generalized Neural Policies for Relational MDPs
• Time-Series Event Prediction with Evolutionary State Graph
• Meta-Reinforcement Learning for Heuristic Planning
• Goal Reasoning by Selecting Subgoals with Deep Q-Learning
• Neural Consciousness Flow
• Understanding Sample Generation Strategies for Learning Heuristic Functions in Classical Planning