Learning Hierarchical Teaching Policies for Cooperative Agents
This addresses scalability issues in multiagent teaching for complex environments, though it is incremental by building on existing action advising approaches.
The paper tackles the problem of scaling learning-to-teach frameworks in cooperative multiagent systems by introducing HMAT, which uses deep representations and hierarchical action advising, resulting in improved team-wide learning progress in complex domains where prior methods fail.
Collective learning can be greatly enhanced when agents effectively exchange knowledge with their peers. In particular, recent work studying agents that learn to teach other teammates has demonstrated that action advising accelerates team-wide learning. However, the prior work has simplified the learning of advising policies by using simple function approximations and only considered advising with primitive (low-level) actions, limiting the scalability of learning and teaching to complex domains. This paper introduces a novel learning-to-teach framework, called hierarchical multiagent teaching (HMAT), that improves scalability to complex environments by using the deep representation for student policies and by advising with more expressive extended action sequences over multiple levels of temporal abstraction. Our empirical evaluations demonstrate that HMAT improves team-wide learning progress in large, complex domains where previous approaches fail. HMAT also learns teaching policies that can effectively transfer knowledge to different teammates with knowledge of different tasks, even when the teammates have heterogeneous action spaces.