Агенты ИИ | AGI_and_RL

Provably Safe Model-Based Meta Reinforcement Learning: An Abstraction-Based Approach https://arxiv.org/abs/2109.01255v1