MCTS — Monte Carlo Tree Search

Algoritma search yang menggabungkan tree search dengan random sampling. Fondasi AlphaGo, AlphaZero, dan game AI modern.

MCTS: 4 langkah - selection, expansion, simulation, backpropagation. Dipopulerkan AlphaGo 2016. Sekarang dipakai di game AI, planning, optimization.

Also known as: Monte Carlo Tree Search
Print

MCTS

Definisi

MCTS (Monte Carlo Tree Search) adalah algoritma search yang menggabungkan tree search dengan random sampling (Monte Carlo).

4 Langkah

  1. Selection — traverse tree dari root dengan UCT/PUCT
  2. Expansion — tambah child node baru
  3. Simulation (rollout) — random play sampai akhir
  4. Backpropagation — update statistik node

Sejarah

  • 2006 — Rémi Coulom (Crazy Stone, Go program) mempopulerkan
  • 2016 — AlphaGo menggunakan MCTS + deep neural networks
  • 2017 — AlphaZero — pure MCTS + neural network
  • 2019+ — MuZero, dll

Aplikasi

  • Game AI (Go, catur, shogi, hex, dll)
  • Planning (robot, logistics)
  • Optimization (combinatorial)
  • Drug discovery (search molecule)
  • Theorem proving (AlphaProof 2024)

Connected to

Not yet written

The following pages are referenced but don't exist yet — they'd make good future additions.

  • /concepts/alphago

References

  1. Wikipedia

Type at least 2 characters to search.

Press to navigate, to open, esc to close.