MuZero — model-based RL without knowing the env dynamics.
hardAnswer
- Schrittwieser et al. 2020: learn a representation + dynamics + reward + value network end-to-end from experience → MCTS in latent space.
- No need for environment simulator (unlike AlphaZero).
- Same NN outputs policy prior + value estimates.
- Trained via combination of TD, MC-return matching, and MCTS policy matching.
- Superhuman Chess/Go/Shogi + top on 57 Atari with a single method.
Check yourself — multiple choice
- Same as AlphaZero
- Learn representation + dynamics + reward + value end-to-end → MCTS in latent space; no env simulator needed; superhuman Chess/Go/Shogi + top on 57 Atari
- Random
- Not real
MuZero: learned dynamics + MCTS in latent; superhuman games + Atari.
#model-based#planning#deep-rl
Practise Reinforcement Learning
214 interview questions in this topic.