MOCHET [7] 
28.07.2018 04:36
 0просмотров 5 0

Evolvestone: An evolutionary generator of balanced digital collectible card games


http://www.sbgames.org/sbgames2016/downloads/anais/156888.pdf

Цитата
The artificial players (bots) were programmed to represent a hu-
man player. There is in the literature [24, 15, 16, 20] several algo-
rithms to implement bots that can compete against human players.
In particular, the Monte Carlo tree search (MCTS) is a classical al-
gorithm that uses a heuristic to analyze the most promising moves
based on an expanding tree that explores only samples of the search
space.   However,  MCTS is still a computing intensive algorithm.
For  this  reason,  in  this  paper,  it  was  developed  a  bot  that  plays
only based on the current turn, making it possible to run faster than
MCTS  and  consequently  more  simulations.   As  both  of  the  bots
play with the same algorithm, the choice of using a simple one will
not have a significant impact in the simulated games because of the
metrics proposed in this paper.

[15] I. Millington. Artificial  intelligence  for  games. Morgan  Kaufmann/Elsevier, Burlington, MA, 2009.
[16] I. Millington. Artificial Intelligence for Games. CRC Press, 2009.
[20] N. Sephton, P. I. Cowling, E. J. Powley, and N. H. Slaven.  Heuristic move pruning in monte carlo tree search for the strategic card game lords of war.  In IEEE Conference on Computational Intelligence and Games (CIG), pages 1–7, 2014.
[24] C. Ward and P. Cowling. Monte carlo search applied to card selection in magic: The gathering.  In IEEE Symposium on Computational Intelligence and Games (CIG), pages 9–16, 2009.




[15] http://lecturer.ukdw.ac.id/~mahas/dossier/gameng_AIFG.pdf
[20] http://orangehelicopter.com/academic/papers/cig2014-low-heuristics.pdf
[24] Monte Carlo search applied to card selection in Magic The Gathering (pdf)



Ensemble Determinization in Monte Carlo Tree Search for the Imperfect Information Card Game Magic: The Gathering
Monte-Carlo Tree Search for Simulation-based Strategy Analysis

Возможность комментировать доступна после регистрации