Dream to Adapt: Meta Reinforcement Learning by Latent Context Imagination and MDP Imagination

Meta reinforcement learning (Meta RL) has been amply explored to quickly learn an unseen task by transferring previously learned knowledge from similar tasks. However, most state-of-the-art Meta RL algorithms require the meta-training tasks to have a dense coverage of the task distribution and a great amount of data for each of them. In this letter, we propose MetaDreamer, a context-based Meta RL algorithm that requires less real training tasks and data by doing meta-imagination and MDP-imagination (Markov-Decision-Process). We perform meta-imagination by interpolating on the learned latent context space with disentangled properties, as well as MDP-imagination through the generative world model where physical knowledge is added to plain VAE networks. Our experiments with various benchmarks show that MetaDreamer outperforms existing approaches in data efficiency and interpolated generalization.