# ASNets: Deep Learning for Generalised Planning

## TL;DR

This paper introduces ASNets, a neural network architecture that learns generalized policies for probabilistic and classical planning problems by exploiting relational structures, enabling quick solutions to larger instances across various domains.

## Contribution

The work extends ASNets to be more expressive and invariant to symmetries, demonstrating their effectiveness through extensive experiments and interpretability analysis.

## Key findings

- ASNets outperform heuristic planners on multiple domains
- ASNets generalize well to large problem instances
- Regularization yields interpretable, compact ASNets

## Abstract

In this paper, we discuss the learning of generalised policies for probabilistic and classical planning problems using Action Schema Networks (ASNets). The ASNet is a neural network architecture that exploits the relational structure of (P)PDDL planning problems to learn a common set of weights that can be applied to any problem in a domain. By mimicking the actions chosen by a traditional, non-learning planner on a handful of small problems in a domain, ASNets are able to learn a generalised reactive policy that can quickly solve much larger instances from the domain. This work extends the ASNet architecture to make it more expressive, while still remaining invariant to a range of symmetries that exist in PPDDL problems. We also present a thorough experimental evaluation of ASNets, including a comparison with heuristic search planners on seven probabilistic and deterministic domains, an extended evaluation on over 18,000 Blocksworld instances, and an ablation study. Finally, we show that sparsity-inducing regularisation can produce ASNets that are compact enough for humans to understand, yielding insights into how the structure of ASNets allows them to generalise across a domain.