> > I'd like to make sure I understand what you mean exactly. You use some > heuristics to intialize all the moves (or maybe some of the moves) with a > certain win-loss and rave-win-loss ratios?
Not only ratios, but also numbers of simulations. Thanks to patterns, expert rules. > > > To a certain extent I suppose these could come from the reading of the > previous move? I think I slowly start to make sense of things... > We have (almost) no "warm start" from the values of previous moves. This should be improved, as very clearly there is some time lost here (e.g. patterns should not be checked from scratch). A+ Olivier
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
