Well...that may the case when you shooting at darkness... the number of eligible moves(top of tree) at any given moment are not that many... probably less than 10... Pro go player filters out eligible moves based on pattern recognition and then they analyze the consequence of each eligible moves... Alphago does similar of filtering based on pattern recognition (trained by neural network learning or whatever) and then runs whatever calculating methods, brute force, Monte carlo...
Slashdot Mirror
Well...that may the case when you shooting at darkness... the number of eligible moves(top of tree) at any given moment are not that many... probably less than 10... Pro go player filters out eligible moves based on pattern recognition and then they analyze the consequence of each eligible moves... Alphago does similar of filtering based on pattern recognition (trained by neural network learning or whatever) and then runs whatever calculating methods, brute force, Monte carlo...