EasyDeepLearn

When does planning beat learning?

medium

Answer

  • (1) Compact known dynamics (games, physics).
  • (2) Long horizons where credit assignment is hard for learning.
  • (3) Rare edge cases (planning can reason zero-shot; learning needs data).
  • (4) Test time compute available.
  • AlphaZero + MuZero show planning + learning outperform either alone.
  • Modern LLM reasoning (o1, DeepSeek-R1) is analogous: test-time 'thinking' as planning augments learned priors.
Check yourself — multiple choice
  • Never
  • Compact known dynamics / long horizons / rare edge cases / test-time compute — planning + learning > either alone (AlphaZero, MuZero); LLM reasoning is analogous
  • Random
  • Always

Planning wins: known dynamics + long horizons + rare cases + test-time compute.

#planning#theory

Practise Reinforcement Learning

214 interview questions in this topic.

Related questions