Back to Glossary
RLReinforcement Learning
強化学習(きょうかがくしゅう)
IntermediateCore Concepts
A machine learning approach where an agent learns to make decisions by receiving rewards or penalties for its actions in an environment.
Why It Matters
RL powers game-playing AI, robotics, and is key to aligning language models with human preferences (RLHF).
Example in Practice
DeepMind's AlphaGo learning to play Go by playing millions of games against itself.