Some Considerations on Learning to Explore via Meta-Reinforcement Learning
This work addresses exploration challenges in meta-reinforcement learning, which is an incremental improvement for researchers in reinforcement learning.
The paper tackled the problem of exploration in meta-reinforcement learning by proposing two new algorithms, E-MAML and E-RL^2, and demonstrated that they deliver better performance on tasks where exploration is important, using a novel 'Krazy World' environment and maze environments.
We consider the problem of exploration in meta reinforcement learning. Two new meta reinforcement learning algorithms are suggested: E-MAML and E-$\text{RL}^2$. Results are presented on a novel environment we call `Krazy World' and a set of maze environments. We show E-MAML and E-$\text{RL}^2$ deliver better performance on tasks where exploration is important.