Re: [Computer-go] Machine for Deep Neural Net training

You can also use hdf5 format, which has transparent compression as well as coffee support. Josef Dne st 27. 4. 2016 18:06 uživatel Gian-Carlo Pascutto napsal: > On 27-04-16 17:45, David Fotland wrote: > > I’d rather just buy another drive than spend time coding and > > debugging another Caffe i

Re: [Computer-go] Is Go group pattern recognition by CNN possible?

Thank you! Dne čt 21. 4. 2016 11:17 uživatel Petr Baudis napsal: > Hi! > > Since "the record's stuck", I have found this as another rant without > point, and djhbrown hasn't responded to my private request for staying > on topic and friendly, I have taken the liberty to "cast the first > sto

Re: [Computer-go] Operators for Frisbee Go Simulation

Hi, I had nothing to do this morning, so I hacked together a very rudimentary version of frisbee-twogtp. The GTP support is done by introducing three new commands: frisbee-play, frisbee-reg_genmove and frisbee-epsilon. I chose to add new commands for play and reg_genmove because of the shift in sem

Re: [Computer-go] Go Bot for the Browser?

If I remember correctly, it is not browser implementation, but rather a frontend. The actual computation runs on server, browser only communicates the moves and shows the results. Josef Dne čt 17. 3. 2016 23:32 uživatel Jim O'Flaherty napsal: > This is wonderful! Tysvm for reposting! > > > On T

Re: [Computer-go] Go Bot for the Browser?

Aha! Thanks for the clarification. Josef Dne pá 18. 3. 2016 9:59 uživatel Darren Cook napsal: > > If I remember correctly, it is not browser implementation, but rather a > > frontend. The actual computation runs on server, browser only > communicates the > > moves and shows the results. > > No,

Re: [Computer-go] AlphaGo & DCNN: Handling long-range dependency

Moreover, it might not be possible to explain the strong play in human understandable terms anyway; human rationalization might simply be a heuristic not strong enough to describe/capture it succinctly. On Mon, Mar 14, 2016 at 3:21 PM Robert Jasiek wrote: > On 14.03.2016 08:59, Jim O'Flaherty wr

Re: [Computer-go] Game 4: a rare insight

How well do you think the mcts-weakness we have witnessed today is hidden in AG? Or, how can one go about exploiting it systematically? I think it might be well hidden by the value network being very strong and true most of the time - it is much harder to get AG to this state, than traditional mct

Re: [Computer-go] AlphaGo won the second game!

Yes, but they are not some random cherry picking third party; have a look on the top authors of the paper - David Silver, Aja Huang, Chris Maddison.. Regards, Josef Dne čt 10. 3. 2016 19:47 uživatel Lukas van de Wiel < lukas.drinkt.t...@gmail.com> napsal: > The same here, with other people havin

Re: [Computer-go] On AlphaGO - confidential

Strictly confidential, I dare say! Dne st 2. 3. 2016 21:04 uživatel "Ingo Althöfer" <3-hirn-ver...@gmx.de> napsal: > Hello Tian, > > the following is confidential, only for you. > > I do not know, how the situation with your go project is. > But please let me know if I can help you in some way or

Re: [Computer-go] Neural Nets to compare human playing strength

By playing data, I mean enough games of Lee Sedol and even less for Shusaku. Dne so 30. 1. 2016 16:07 uživatel Josef Moudrik napsal: > 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 > stren

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 t

Re: [Computer-go] Go Aesthetics

And do you find these "ugly yet working" moves aesthetically pleasing? I think it all depends what do we mean by aesthetics. In my opinion, it is not strength - the hard thing about go imo is that while the nice (shape, ..) do often work, sometimes, the ugly move works better - precisely as Nick w

Re: [Computer-go] Go Aesthetics

Some time ago, we (with pasky) looked into a related question of player attribute prediction - such as territoriality, aggresivity, influence oriented style, or strength. Project website is here: http://gostyle.j2m.cz/ Imo you can predict/evaluate pretty much anything you get dataset for. Aestheti

Re: [Computer-go] CNN with 54% prediction on KGS 6d+ data

Hi, Winrate of your pure CNN againts pachi retsugen is: GAMES WINRATE S.D.PAIRING 224 0.558 0.033 19-7.5-1-pachi-=1-detlef_54 221 0.407 0.033 19-7.5-1-pachi-=2-detlef_54 I used the https://github.com/jmoudrik/deep-go-wrap for the player. Regards, Josef On Tue, Dec 29, 2

Re: [Computer-go] CNN with 54% prediction on KGS 6d+ data

would have to be > executed on the GPU, with no round-trips to the CPU. That's what my > email was aimed at. > > On Tue, Dec 08, 2015 at 04:37:05PM +, Josef Moudrik wrote: > > Regarding full CNN playouts, I think that problem is that a playout is a > > long seri

Re: [Computer-go] CNN with 54% prediction on KGS 6d+ data

Regarding full CNN playouts, I think that problem is that a playout is a long serial process, given 200-300 moves a game. You need to construct planes and transfer them to GPU for each move and read result back (at least with current CNN implementations afaik), so my guess would be that such playou

Re: [Computer-go] Strong engine that maximizes score

> > I am trying to create a database of games to do some machine-learning > experiments. My requirements are: > * that all games be played by the same strong engine on both sides, > * that all games be played to the bitter end (so everything on the board > is alive at the end), and > * that both

Re: [Computer-go] Frisbee Go Simulation

> > (effectively) zero. Therefore epsilon should be pretty tiny. It must be > large enough that there is a chance of the frisbee being at least 50% over > the line (i.e. epsilon > 0), but small enough that the chance of it going > 70.7% over the line is vanishingly small (otherwise we would be allo

Re: [Computer-go] Standard Computer Go Datasets - Proposal

On Fri, Nov 13, 2015 at 11:16 AM Erik van der Werf wrote: > On Fri, Nov 13, 2015 at 10:46 AM, Darren Cook wrote: >> >> The advantages of storing games: >> * accountability/traceability >> * for programs who want to learn sequences of moves. >> > > Another advantage of storing games is that i

Re: [Computer-go] Standard Computer Go Datasets - Proposal

Hello, On Fri, Nov 13, 2015 at 10:13 AM wrote: > I would only use it if it is licensed for commercial use. Yes, I would like to licence this as such, please see below. On Fri, Nov 13, 2015 at 10:23 AM Petr Baudis wrote: > I think the current de facto standard dataset is GoGoD (some year, not

[Computer-go] Standard Computer Go Datasets - Proposal

2000 games?) and large dataset (5-6 games). 2b) Strength & year span: Currently I am thinking about including modern professional games only (1970-2015) 3) Do you have any other comments, requirements for the dataset and ideas? Thanks for Your attention, Kind regards

Re: [Computer-go] Scraping lower-ranked games from kgs

Hello, I have a simple script for downloading info (gamelist, list of oponents, games) one by one here: http://repo.or.cz/gostyle.git/blob/HEAD:/kgs/kgs.py If you want to download more players, you need to search through opponents of players you know. But do not spam the kgs server too much plea

Re: [Computer-go] Frisbee Go Simulation

Frisbee go sounds fun. How do you plan to use the GTP protocol to support this? I think that the randomization should be handled by the server, so the bot needs to get feedback about the move actually carried out. So maybe genmove + undo & play or reg_genmove + play depending on what do the bots su

Re: [Computer-go] Neural Network Chess Computer Abandons Brute Force For "Human" Approach

Actually, I opened the paper this morning (because of the sensational headline on slashdot), and I do not find it very interesting; the author only uses nn to train the evaluation function (then uses search tree) and there seems to be nothing new about this at all... Josef Dne út 15. 9. 2015 5:15

Re: [Computer-go] CGT endgame solver

Hello, thanks for the replies everyone. Regards, Josef Dne so 18. 7. 2015 1:59 uživatel Igor Polyakov napsal: > I'm working on Iomrascalai with Urban Hafner. It's fairly strong at 9x9 > (better than GnuGo), but needs work on larger boards. > > On 2015-07-14 0:22, "Ingo Althöfer" wrote: > > Hi,

[Computer-go] CGT endgame solver

Hello! As far as I know, combinatorial game theory is not used in modern Go engines, despite its nice theoretical properties. I have been wondering why this is so. Or am I mistaken and some engines do use it? In general, I only know about Martin Mueller's Explorer program, which is however slightly

[Computer-go] Computer Go Turnament at EGC 2015: Call for Participation

://pasky.or.cz/iggsc2015/compgo_spec.html Looking forward to Your participation, Kind regards, Josef Moudrik ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Standalone DNN player support

f On Thu, Apr 30, 2015 at 12:25 PM Josef Moudrik wrote: > > I would love to have something like this. > Great! > > > I would appreciate some way to configure depth levels and > > variable branching factors for move generation as well as scoring > > playouts using the N

[Computer-go] Extended Deadline: Second International Go Game Science Conference, Liberec 2015

Hello! This is a notification that we have extended the paper submission deadline for the IGGSC 2015 at the European Go Congress to May 25, 2015. Call for Papers - Second International Go Game Science Conference = Liberec, Czech Rep

Re: [Computer-go] topological go & gtp

GTP is straightforward, have a look at michi for a simple implementation example.. Josef On Wed, May 13, 2015 at 11:55 PM Ray Tayek wrote: > On 5/7/2015 9:22 AM, Ray Tayek wrote: > > hi, i am writing a go program in java that will let you play on a torus > like little golem >

Re: [Computer-go] Standalone DNN player support

> I would love to have something like this. Great! > I would appreciate some way to configure depth levels and > variable branching factors for move generation as well as scoring > playouts using the NN. Hmm, I am not talking about MCTS integration or DNN training. Rather, I have a small wrapper

[Computer-go] Standalone DNN player support

Hi! I am playing around with DNN move prediction (as everyone here seems to be doing at the moment) and I wonder what tools do you use to wrap the net to play correctly & support GTP. If there is no such wrapper, I am planning to write it myself (which should be fairly easy). I would like to hear

Re: [Computer-go] [ANN] Imago - Go board optical recognition

Hello Andrea, I happen to be organizing the "technical section" of EGC2015, and we would certainly be really glad to use automatic transcription and internet broadcasting, as this would make it possible to broadcast more boards with less manpower (hopefully more precisely, it is common that one hu

Re: [Computer-go] CFP: Second International Go Game Science Conference, Liberec 2015

Hello Ingo, Yes, both events are in the first week. The tournament is scheduled for 29th July, conference the day after. A part of the conference might be on 29th as well, depending on the number of presentations. Kind regards, Josef Dne 13. 12. 2014 14:25 Ingo Althöfer <3-hirn-ver...@gmx.de> nap