Learning Probabilistic Relational Dynamics for Multiple Tasks
This work addresses the challenge of data-efficient learning of action dynamics for agents in relational domains, though it appears incremental as it builds on existing rule-based and transfer learning methods.
The paper tackles the problem of learning relational probabilistic planning rules for multiple related tasks by developing a hierarchical Bayesian approach with a prior distribution over rule sets, showing that transferring information between tasks significantly reduces the training data needed to predict action effects in blocks-world domains.
The ways in which an agent's actions affect the world can often be modeled compactly using a set of relational probabilistic planning rules. This paper addresses the problem of learning such rule sets for multiple related tasks. We take a hierarchical Bayesian approach, in which the system learns a prior distribution over rule sets. We present a class of prior distributions parameterized by a rule set prototype that is stochastically modified to produce a task-specific rule set. We also describe a coordinate ascent algorithm that iteratively optimizes the task-specific rule sets and the prior distribution. Experiments using this algorithm show that transferring information from related tasks significantly reduces the amount of training data required to predict action effects in blocks-world domains.