I'm having trouble beating an AI in Isolation game: https://en.wikipedia.org/wiki/Isolation_(board_game) with 3 queens on a 7x7 board. I tried applying alpha beta iteration with a scoring function on the state. I tried 3 scoring functions:

  1. (0.75/# moves taken) * number of legal moves - (# moves taken/0.75) * number of opponent legal moves
  2. number of legal moves - 3 * number of opponent legal moves
  3. 3 * number of legal moves - number of opponent legal moves

The first is an annealed aggressive strategy so the agent gets more aggressive as the game goes longer. The 2nd is a pure aggression strat and the last is a pure defensive strat. None of them consistently beat standard alpha beta iteration with the state scoring function: number of legal moves - number of opponent legal moves. They all broke roughly even.

Any suggestions of scoring state functions or search algorithms are appreciated. The search has a limit of 6000 seconds per turn though.

  • $\begingroup$ What is your main specific question here? Please, choose one specific question and put it in the body and title of your post to clarify what you're asking here. $\endgroup$
    – nbro
    Feb 8 '21 at 10:53
  • $\begingroup$ I'm simply asking for a search algorithm with a cost function that can do better than iterative alpha beta $\endgroup$ Feb 8 '21 at 16:56
  • 1
    $\begingroup$ Is your question somehow related to this: ai.stackexchange.com/q/5174/2444? $\endgroup$
    – nbro
    Feb 9 '21 at 17:31

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.