Learning Interpretable Classifiers for PDDL Planning

TL;DR

This paper introduces a method for learning interpretable, human-readable logical formulas that describe agent behavior in PDDL planning tasks, enabling understanding and generalization of policies.

Contribution

It presents a novel topology-guided MaxSAT approach to efficiently learn interpretable behavior classifiers from planning examples.

Findings

Formulas are human-readable and generalize to unseen instances.

Learning is computationally intractable without approximation.

The MaxSAT-based method produces accurate formulas in reasonable time.

Abstract

We consider the problem of synthesizing interpretable models that recognize the behaviour of an agent compared to other agents, on a whole set of similar planning tasks expressed in PDDL. Our approach consists in learning logical formulas, from a small set of examples that show how an agent solved small planning instances. These formulas are expressed in a version of First-Order Temporal Logic (FTL) tailored to our planning formalism. Such formulas are human-readable, serve as (partial) descriptions of an agent's policy, and generalize to unseen instances. We show that learning such formulas is computationally intractable, as it is an NP-hard problem. As such, we propose to learn these behaviour classifiers through a topology-guided compilation to MaxSAT, which allows us to generate a wide range of different formulas. Experiments show that interesting and accurate formulas can be learned in reasonable time.