Mastering the game of Go with deep neural networks and tree search¶
Silver, D., Huang, Maddison, Guez, Sifre, den Driessche, v., Schrittwieser, et al. (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529(7587), 484-489.
Cited by¶
1 citation across 1 artifact.
Primes¶
- Markov Decision Processes (MDPs)
- The solution approach, as detailed by Silver et al. (2016), combines: Monte Carlo Tree Search (MCTS) as the decision-time planning algorithm, producing action selection by exploring the game tree with selective deepening guided by value estimates; deep-neural-network value and policy estimates that guide MCTS (the policy network biases action selection toward likely-good moves; the value network estimates state values without full tree expansion); and reinforcement learning from self-play that trains the value and policy networks through iterated games between successive agent versions, with the training signal being actual game outcomes combined with MCTS improvement over raw policy outputs.
Supported in partVerified against the publisher's abstract
The AlphaGo abstract states the search combines Monte Carlo simulation with value and policy networks, trained by supervised learning plus reinforcement learning from self-play.
“We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks.”
- The solution approach, as detailed by Silver et al. (2016), combines: Monte Carlo Tree Search (MCTS) as the decision-time planning algorithm, producing action selection by exploring the game tree with selective deepening guided by value estimates; deep-neural-network value and policy estimates that guide MCTS (the policy network biases action selection toward likely-good moves; the value network estimates state values without full tree expansion); and reinforcement learning from self-play that trains the value and policy networks through iterated games between successive agent versions, with the training signal being actual game outcomes combined with MCTS improvement over raw policy outputs.
Verification¶
Does it exist? Not checked yet. This entry carries no identifier to resolve. It was extracted from the citation as written in the article, normalized, and deduplicated against the rest of the registry.
Does it back the claim? Read against the text for 1 of 1 citation: 1 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.
Was it audited? Yes. A second, independent pass read the citation against the article text and recorded a verdict.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:cde8b83d4f8e · see in the full table