Re: [computer-go] Paper: Beta Distribution
Peter, I tried to reproduce this, so I gave this a whirl and the win rate against UCB-Tuned1 with first move priority of 1.1 (like Mogo) was only 33%. That was using uniform random playouts. What was the playout policy you used for this? Christian On 18/06/2009 21:04, Peter Drake wrote: An improvement on the UCB/UCT formula: Stogin, J., Chen, Y.-P., Drake, P., and Pellegrino, S. (2009) “The Beta Distribution in the UCB Algorithm Applied to Monte-Carlo Go”. In Proceedings of the 2009 International Conference on Artificial Intelligence, CSREA Press. http://webdisk.lclark.edu/drake/publications/BetaDistribution.pdf Peter Drake http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Paper: Beta Distribution
I believe we used a uniform random policy (only don't play in your own pseudoeyes). The numbers probably won't be the same, but we've certainly replicated the qualitative improvement with version 6.05 of Orego, available here: https://webdisk.lclark.edu/drake/orego/ Peter Drake http://www.lclark.edu/~drake/ On Jun 23, 2009, at 9:24 AM, Christian Nentwich wrote: Peter, I tried to reproduce this, so I gave this a whirl and the win rate against UCB-Tuned1 with first move priority of 1.1 (like Mogo) was only 33%. That was using uniform random playouts. What was the playout policy you used for this? Christian On 18/06/2009 21:04, Peter Drake wrote: An improvement on the UCB/UCT formula: Stogin, J., Chen, Y.-P., Drake, P., and Pellegrino, S. (2009) “The Beta Distribution in the UCB Algorithm Applied to Monte-Carlo Go”. In Proceedings of the 2009 International Conference on Artificial Intelligence, CSREA Press. http://webdisk.lclark.edu/drake/publications/BetaDistribution.pdf Peter Drake http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Paper: Beta Distribution
An improvement on the UCB/UCT formula: Stogin, J., Chen, Y.-P., Drake, P., and Pellegrino, S. (2009) “The Beta Distribution in the UCB Algorithm Applied to Monte-Carlo Go”. In Proceedings of the 2009 International Conference on Artificial Intelligence, CSREA Press. http://webdisk.lclark.edu/drake/publications/BetaDistribution.pdf Peter Drake http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Paper: Beta Distribution
Section 3.2 describes a pair of tests that took about 4.2 minutes each (if my calculations are correct). Why not play more games and have each game contain more simulations? Writing the code and the paper is the hard part, waiting for a computer to run your code is easy. Peter Drake wrote: An improvement on the UCB/UCT formula: Stogin, J., Chen, Y.-P., Drake, P., and Pellegrino, S. (2009) “The Beta Distribution in the UCB Algorithm Applied to Monte-Carlo Go”. In Proceedings of the 2009 International Conference on Artificial Intelligence, CSREA Press. http://webdisk.lclark.edu/drake/publications/BetaDistribution.pdf Peter Drake http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Paper: Beta Distribution
On Thu, Jun 18, 2009 at 6:43 PM, Michael Williamsmichaelwilliam...@gmail.com wrote: Section 3.2 describes a pair of tests that took about 4.2 minutes each (if my calculations are correct). Why not play more games and have each game contain more simulations? Writing the code and the paper is the hard part, waiting for a computer to run your code is easy. I guess there is a mistake in that paragraph, since you can't run a full game in .25 seconds. That's probably the time per move, so the whole test might have taken 10 or 20 hours. Peter Drake wrote: An improvement on the UCB/UCT formula: Stogin, J., Chen, Y.-P., Drake, P., and Pellegrino, S. (2009) “The Beta Distribution in the UCB Algorithm Applied to Monte-Carlo Go”. In Proceedings of the 2009 International Conference on Artificial Intelligence, CSREA Press. http://webdisk.lclark.edu/drake/publications/BetaDistribution.pdf Peter Drake http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/