artificial intelligence - What to Do when Monte Carlo Tree Search Hits Memory Limit -
i have taken interest monte carlo tree search applied in games recently.
i have read several papers, use "monte-carlo tree search" phd thesis chaslot, g find more easy understand basics of monte carlo tree search
i have tried code it, , stuck on problem. algorithm tries expand 1 node game tree every 1 simulation. escalates memory problem. have read paper, doesnt seem explain technique if hits memory limit.
can suggest should technique if hits memory limit?
you can see paper here : http://www.unimaas.nl/games/files/phd/chaslot_thesis.pdf
you can throw away nodes number of visits smaller threshold not visited (how many playouts ago). that's quick not efficient solution. it's better implement progressive widening too.
Comments
Post a Comment