Reinforcement Learning Algorithms with Python

Understanding ME-TRPO

In the first part of ME-TRPO, the dynamics of the environment (that is, the ensemble of models) are learned. The algorithm starts by interacting with the environment with a random policy, , to collect a dataset of transitions, . This dataset is then used to train all the dynamic models, , in a supervised fashion. The models, , are initialized with different random weights and are trained with different mini-batches. To avoid ...