Learning Lifted STRIPS Models from Action Traces Alone: A Simple, General, and Scalable Solution

2024-11-22

arXiv.org

AI Summary

This paper addresses the problem of automatically learning lifted STRIPS action models from raw action trajectories—without predefined predicates. To tackle the challenges of implicit predicate discovery and joint inference of effect patterns and preconditions, we propose the first general learning framework that simultaneously achieves scalability (comparable to LOCM), logical completeness, and soundness (akin to SAT-based methods). Our approach introduces a novel trajectory-driven effect consistency checking mechanism, supporting predicates of arbitrary arity and unrestricted hidden domains. It integrates lifted representation learning, logical modeling, and satisfiability verification. Evaluated on large-scale trajectories—exceeding hundreds of thousands of state transitions—in classic domains such as the 8-puzzle, our method successfully learns compact and accurate STRIPS models. Furthermore, we demonstrate strong generalization capability on significantly larger problem instances.

Technology Category

• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Learning STRIPS action models from action traces alone is a challenging problem as it involves learning the domain predicates as well. In this work, a novel approach is introduced which, like the well-known LOCM systems, is scalable, but like SAT approaches, is sound and complete. Furthermore, the approach is general and imposes no restrictions on the hidden domain or the number or arity of the predicates. The new learning method is based on an emph{efficient, novel test} that checks whether the assumption that a predicate is affected by a set of action patterns, namely, actions with specific argument positions, is consistent with the traces. The predicates and action patterns that pass the test provide the basis for the learned domain that is then easily completed with preconditions and static predicates. The new method is studied theoretically and experimentally. For the latter, the method is evaluated on traces and graphs obtained from standard classical domains like the 8-puzzle, which involve hundreds of thousands of states and transitions. The learned representations are then verified on larger instances.

Problem

• Learning STRIPS models from action traces without domain predicates
• Ensuring scalability, soundness, and completeness in model learning
• Generalizing across hidden domains with no predicate restrictions

Innovation

• Efficient novel test for predicate consistency
• Scalable and general STRIPS model learning
• Sound and complete like SAT approaches