Learning Non-Deterministic Multi-Agent Planning Domains

In this paper, we present an algorithm for learning non- deterministic multi-agent planning domains from execution examples. The algorithm uses a master-slave decomposi- tion of two population-based stochastic local search algo- rithms and integrates binary decision diagrams to reduce th e size of the search space. Our experimental results show that the learner has high convergence rates due to an aggressive exploitation of example-driven search and an efficient sep- aration of concurrent activities. Moreover, even though th e learning problem is at least as hard as learning disjoint DNF formulas, large domains can be learned accurately within a few minutes.