I have these Monte Carlo Tree Search, and I need to expand it, but I don't understand the step 1 and 2, why it it goes to the first Node and then make a new Node? instead of go to the depthest left Leaf?

I thought it need to go to the Most Probability Leaf.

MonteCarlo Tree Search: enter image description here

Monte Carlo Tree Step 1, it added a New Node, now are 9 Cases: enter image description here Monte Carlo Tree Step 2, it added a New Node, now are 2 Cases under the First Node:enter image description here

  • $\begingroup$ The question reads a bit messy, so let us sort the facts. According to the OP the game is called TicTacToe. The screenshot shows a part of the gametree. This indicates, that the game engine, better known as the move generator, is working. That means, the rules of TicTacToe are implemented and the game engine can create a random move. $\endgroup$ – Manuel Rodriguez Feb 24 at 16:38
  • $\begingroup$ ... Now the aim is to use the existing forward model together with a solver to find the winning strategy. This is done with Monte Carlo Tree search (MCTS), which is a powerful technique for testing out different action sequences in an existing game engine. If i understand the problem correct, something is wrong with the MCTS-solver and it's not possible to output the correct next move. Or to be more specific, it has to do with which node in the gametree should be expanded next. $\endgroup$ – Manuel Rodriguez Feb 24 at 16:39
  • $\begingroup$ @ManuelRodriguez my Question is, could I take the last Leaf instad to create a new Node , instead of create a new Node into the 2 Node? $\endgroup$ – TomaateTip Feb 24 at 17:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.