Seonglae ChoModel-Based
Reinforcement learning models often fail to learn when there are even small changes in the environment, so [we want them] to adapt well like zero shot generalization
to solve contextual RL problems with contextual
loss by contextual similarity within with choosing optimal training environment based on for steady and fast training.
Model-Based Transfer Learning for Contextual Reinforcement Learning
https://jhoon-cho.github.io/MBTL/
arxiv.org
https://arxiv.org/pdf/2408.04498