Learning Symbolic Models of Stochastic Domains

Abstract

In this article, we work towards the goal of developing agents that can learn to act in complex worlds. We develop a probabilistic, relational planning rule representation that compactly models noisy, nondeterministic action effects, and show how such rules can be effectively learned. Through experiments in simple planning domains and a 3D simulated blocks world with realistic physics, we demonstrate that this learning algorithm allows agents to effectively model world dynamics.

References in corpus (5)

• Tractable Inference for Complex Stochastic Processes
• Learning Partially Observable Deterministic Action Models
• Logical Hidden Markov Models
• Inductive Policy Selection for First-Order MDPs
• Relational Dynamic Bayesian Networks

Cited by in corpus (21)

• Symbol Emergence in Cognitive Developmental Systems: a Survey
• Learning STRIPS Operators from Noisy and Incomplete Observations
• A Survey of Optimization-based Task and Motion Planning: From Classical To Learning Approaches
• From Machine Learning to Robotics: Challenges and Opportunities for Embodied Intelligence
• Human-Level Reinforcement Learning through Theory-Based Modeling, Exploration, and Planning
• Deliberative Acting, Online Planning and Learning with Hierarchical Operational Models
• PDDLGym: Gym Environments from PDDL Problems
• Adaptable and Verifiable BDI Reasoning
• Exploring compact reinforcement-learning representations with linear regression
• Symbolic Manipulation Planning with Discovered Object and Relational Predicates
• Deep Affordance Foresight: Planning Through What Can Be Done in the Future
• Learning Action Models from Disordered and Noisy Plan Traces
• Learning Type-Generalized Actions for Symbolic Planning
• Neuro-Symbolic Imitation Learning: Discovering Symbolic Abstractions for Skill Learning
• Representation Learning for Classical Planning from Partially Observed Traces
• Relational Learning for Skill Preconditions
• Learning sparse relational transition models
• Evaluating the Apperception Engine
• Incremental Learning of Discrete Planning Domains from Continuous Perceptions
• Metric Learning for Generalizing Spatial Relations to New Objects
• Learning Numerical Action Models from Noisy Input Data