Die wichtigste Info vorab: Die Komodo Dragon 2 Engine kommt an die Spielstärke einer kostenlosen Stockfish 13 Engine nicht heran. AlphaGo doesn't use minimax with or without pruning, it uses MCTS (with pruning). Komodo Dragon 2 ist die aktuellste Entwicklungsstufe der Komodo-Engine und will mit dem Untertitel Neural Network Chess Evaluation auf den AI-Trend aufspringen. That has nothing to do with minimax though. It's right in the name! See for example the first paragraph on page 20 of, which clearly describes AG constructing a search tree. >You described the "Monte Carlo" part of MCTS but not the "Tree Search" part. But it can run in time constrained systems better because it picks a much smarter order to run in. At t=infinity, MCTS is equivalent to Minimax, because it will try all nodes. MCTS uses monte-carlo methods to estimate the best paths instead of trying all of them naively. To be clear, what I meant, that you pulled out of context is that Minimax requires an exact calculation of the value of each child before choosing the best and returning it, this requires either simulating every possible game, or an evaluation function (depth-pruning). It uses a static evaluation function to evaluate positions without searching down the tree furtherĮhh alright, it calculates the value of a node exactly (or exactly assuming its evaluation function). >No, I assure you that stockfish definitely does not look "all the way down (ie to states where the game is over)" in most positions, even with pruning - that would take too long. This is true in both MCTS and in Stockfish. When we truncate it we use an approximation of the value of the node (the static evaluation function) when we prune we sometimes use the static evaluation function as well. Hence we prune and truncate the search tree. Minimax obviously will produce perfect play in both chess and Go, but we can't use it because it takes too long. You described the "Monte Carlo" part of MCTS but not the "Tree Search" part. > MCTS simulates games randomly and creates a distribution of expected value for each one based on some cool math. It uses a static evaluation function to evaluate positions without searching down the tree further. No, I assure you that stockfish definitely does not look "all the way down (ie to states where the game is over)" in most positions, even with pruning - that would take too long. Minimax actually needs to look at all possible branches all the way down. No, stockfish uses minimax with pruning.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |