Reinforcement Learning
Reinforcement learning is a machine-learning paradigm in which an agent learns a policy through interaction so as to improve long-term cumulative reward.
Definition
Reinforcement learning is a machine-learning paradigm in which an agent learns a policy through interaction so as to improve long-term cumulative reward.
Technical Context
State, action, reward, policy, and value are central concepts. If the reward poorly represents the real objective, specification gaming or reward hacking can emerge.