Reinforcement Learning

noun

say it: “ree-in-FORS-ment LUR-ning” /ˌriɪnˈfɔrsmənt ˈlɝnɪŋ/

A kind of machine learning in which a program learns which actions to take by trying them and being rewarded or penalized for the results.

What is reinforcement learning?

Reinforcement learning is machine learning where a program learns by doing: it tries an action, gets a reward or a penalty, and over many tries learns which actions pay off, like a game player that gets better by playing thousands of games. Nobody gives it the right answers.

In a sentence

  • “In reinforcement learning, the score after each game is the only feedback the program gets.”
  • “The robot learned to walk through reinforcement learning, falling thousands of times.”

Level: Journeyman · More AI Basics terms at this level

Updated