Meta-Learning
Meta-learning can be described as 'learning to learn' by training on various tasks.
One example is MAML (Model-Agnostic Meta-Learning), which learns an initial weight.
It consists of an inner loop and an outer loop.
The inner loop learns a specific task, while the outer loop updates the initial weights by looping through the inner loops.
Meta-RL
Meta-RL is the task of applying Meta-Learning to RL.
I glanced recent paper, Meta-RL can be tested with mujoco.
I think I can research too.
Other field's research need too much computer power..
multi-task RL
I will add more information after be more smarter.
