Hi Josef, thanks for the links to your interesting projects.
In general, I think CNNs are sort of our new hammer. We should walk around (in our Go universe) and test the hammer on all possible questions... Perhaps we have reached iron age now, after (Crazy)Stone age... Ingo. Gesendet: Samstag, 30. Januar 2016 um 16:07 Uhr Von: "Josef Moudrik" <j.moud...@gmail.com> An: computer-go <computer-go@computer-go.org> Betreff: Re: [Computer-go] Neural Nets to compare human playing strength Imo, you do not have enough playing data to train full network. One way to do this might be to have a general system to estimate player strength from his records. My http://gostyle.j2m.cz project does something similar, but for strong players it is not very precise.. I have recently tried to do this using cnn (from position, predict rank of both players) and it works a bit, but still with high error and variance, so it is not very reliable. See my blog for a report on this http://jmoudrik.github.io/post/2016/01/15/convolutional_neural_net_for_Go_strength_prediction.html[http://jmoudrik.github.io/post/2016/01/15/convolutional_neural_net_for_Go_strength_prediction.html] Regards, Josef Dne so 30. 1. 2016 15:50 uživatel Petri Pitkanen <petri.t.pitka...@gmail.com[petri.t.pitka...@gmail.com]> napsal: I do not think such exercise would give any meaningful results. NN would not imitate it's 'hero' 1-1 not even close Funny such discussion keep on going on in Chess and Go, in chess i think Steiniz would be wiped of board by best players of today, but still he would be way better as he created much of modern chess anyway. I am pretty sure such players exists in Go as well 2016-01-30 14:23 GMT+02:00 "Ingo Althöfer" <3-hirn-ver...@gmx.de[3-hirn-ver...@gmx.de]>:Hi, in a German computer-go subforum we have a discussion which involves a comparison of the "absolute" playing strengths of current hero Lee Sedol and Shusaku (1829-1862). http://senseis.xmp.net/?Shusaku[http://senseis.xmp.net/?Shusaku] It is not possible to let them play a match against each other. But what about the following: Build two neural nets A and B with the same overall structure. Let A be trained with the games of Sedol, and B be trained with the games of Shusaku. Then let go bots based on A and B play against each other. Would the outcome typically say something about the playing strengths of Sedol and Shusaku in comparison? Ingo. _______________________________________________ Computer-go mailing list Computer-go@computer-go.org[Computer-go@computer-go.org] http://computer-go.org/mailman/listinfo/computer-go_______________________________________________ Computer-go mailing list Computer-go@computer-go.org[Computer-go@computer-go.org] http://computer-go.org/mailman/listinfo/computer-go_______________________________________________ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go[http://computer-go.org/mailman/listinfo/computer-go] _______________________________________________ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
