Diverse Policy Optimization for Structured Action Space
This addresses the challenge of scaling diverse policy discovery in RL for tasks with complex structured actions, which is incremental by building on existing probabilistic RL and GFlowNet techniques.
The paper tackled the problem of enhancing policy diversity in reinforcement learning with structured action spaces, proposing Diverse Policy Optimization (DPO) which uses energy-based models and GFlowNet for efficient sampling, and demonstrated substantial outperformance over state-of-the-art methods on ATSC and Battle benchmarks.
Enhancing the diversity of policies is beneficial for robustness, exploration, and transfer in reinforcement learning (RL). In this paper, we aim to seek diverse policies in an under-explored setting, namely RL tasks with structured action spaces with the two properties of composability and local dependencies. The complex action structure, non-uniform reward landscape, and subtle hyperparameter tuning due to the properties of structured actions prevent existing approaches from scaling well. We propose a simple and effective RL method, Diverse Policy Optimization (DPO), to model the policies in structured action space as the energy-based models (EBM) by following the probabilistic RL framework. A recently proposed novel and powerful generative model, GFlowNet, is introduced as the efficient, diverse EBM-based policy sampler. DPO follows a joint optimization framework: the outer layer uses the diverse policies sampled by the GFlowNet to update the EBM-based policies, which supports the GFlowNet training in the inner layer. Experiments on ATSC and Battle benchmarks demonstrate that DPO can efficiently discover surprisingly diverse policies in challenging scenarios and substantially outperform existing state-of-the-art methods.