Re: [Computer-go] AMAF/RAVE + heavy playouts - is it save?

[Robert Finking]

  
  
You are welcome. Figure 1 in [2] is the diagram I was thinking
  of. 

On 03-Nov-15 20:39, Tobias Pfeiffer
  wrote:


  
  This helps very much, thank you for taking the time to answer!
  
  You might be looking for for "Combining Online and Offline
  Knowledge in UCT" [1] by Gelly and Silver. Silver Tesauroreference
  it in "Monte-carlo Simulation Balancing" [2] with "Unfortunately,
  a stronger simulation policy can actually lead to a weaker
  Monte-Carlo search (Gelly & Silver,
  2007), a paradox that we explore further in this paper."
  
  I'll make it a priority to read both papers in detail thank you!
  If you meant another paper, someone else knows one I'm happy to
  see more references.
  
  Thanks!
  Tobi
  
  
  [1] http://www.machinelearning.org/proceedings/icml2007/papers/387.pdf
  [2] http://www.machinelearning.org/archive/icml2009/papers/500.pdf
  
  
  On 03.11.2015 21:03, robertfinkng...@o2.co.uk
wrote:
  
  

You have to be careful what heuristics you apply. This was a
  surprising result: using a playout policy which in itself is a
  stronger go player can actually make MCTS/AMAF weaker. The
  reason is that MCTS depends entirely on accurate estimations
  of the value of each position in the tree. Any playout policy
  which introduces a bias therefore weakens MCTS. It may
  increase precision (lower standard deviation) but gives a less
  accurate assessment of the value (an incorrect mean). Most
  playouts at the moment (at least published ones) are based on
  Remi's Mogo playout policy, which increases precision without
  sacrificing accuracy.
  
  There's a really nice diagram in one of David Silver's papers
  illustrating the effect that bias can have on playouts. As
  soon as you see it you understand the problem. Unfortunately I
  don't have it to hand and have unfortunately run out of time
  looking for it, otherwise I'd reference it. Hopefully somebody
  else can give the reference. I suspect David probably
  co-authored the paper in which case apologies to the other
  author for not crediting them here!
  
  I hope this helps
  
  Regards
  
  Raffles

On 03-Nov-15 19:38, Tobias Pfeiffer
  wrote:


  Hi everyone,

I haven't yet caught up on most recent go papers. If what I ask is
answered in one of these, please point there.

It seems everyone is using quite heavy playouts these days (nxn
patterns, atari escapes, opening libraris, lots of stuff that I don't
know yet, ...) - my question is how does that mix with AMAF/RAVE? I
remember from the early papers, that they said it'd be dangerous to do
it with non random playouts and that they shouldn't have too much logic.

Which, well, makes sense (to me) because the argument is that we play
random moves so they are order independent. With patterns that doesn't
hold true anymore.

What's the experience out there? Does it just still work? Does it not
matter because you just "warm up" the tree? Or do you need to be careful
with what heuristics you apply not too break RAVE/AMAF?

Thank you!
Tobi


  
  
  
  ___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go
  
  
  
  -
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7163 / Virus Database: 4457/10906 - Release Date: 10/28/15






___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go
  
  
  -- 
www.pragtob.info


  

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Justin Blank]

I have repeatedly seen people assert that komi must be different for
players of different skill levels, and have repeatedly questioned it, but I
have never seen anyone try to substantiate the claim. People who believe it
find it obvious, but I don't. There are two pieces of evidence that I can
think of:

1) that I believe someone played near-random engines against each other,
and the correct komi was different (I cannot find where that was done). But
that's so far from even DDK play that it's pretty useless.
2) I believe the old OGS (DGS?) forums included an analysis of their games
and what the correct komi was. I cannot confidently quote the results. If
those are the old OGS forums, I don't know if they even exist anymore.

The data from go servers are freely available. Does white have a greater
advantage for weaker players? It doesn't seem so--anecdotally when players
posted their KGS stats, they varied a bit, but didn't seem to have a bias
for White.

Of course, that's anecdata...anyone is welcome to prove or disprove this
old claim by analyzing the stats on KGS, or Tygem or wherever else.

On Thu, Nov 5, 2015 at 2:03 AM, Petri Pitkanen 
wrote:

> Let alone we do not have even sufficient understanding of perfect play to
> say what is correct komi in absolute sense. Nor it is it even meaningful
> concept. Correct komi is a komi that produces about 50/50 result. Obviously
> komi that will result in 50/50 for professionals will probably favour white
> in your average weekend tournaments. Just like in chess 1st move advantage
> is clearly less meanigfull for weaker players than top professionals.
>
> So setting komi is not theroretical but statistical issue
>
> 2015-11-05 0:04 GMT+02:00 Hideki Kato :
>
>> The correct komi value assuming both players are perfect.  Or, black
>> utilize his advantage (maybe in an early stage) perfectly.  Actual
>> players, even strong pros, are not perfect and cannot fully utilize
>> their advantages.  As a conclusion, white is favored.
>>
>> Hideki
>>
>> Aja Huang: 

Re: [Computer-go] Komi 6.5/7.5

[Nick Wedd]

Many years ago, when "auction komi" was tried out at a London Open Go
tournament, I collected statistics on what the winning komi bid was among
players of various strengths. There was a positive correlation between
playing strength and komi.  Of course this does not answer the question, it
just shows that strong humans valued the right to play first more than weak
humans did.

I find Hideki Kato's argument intuitively convincing. Suppose "God's komi"
is *n*, so a game between two perfect players using *n* komi ends in jigo.
But when two imperfect players use *n* komi, they will make imperfect
moves, which will reduce the value of sente, and of komi, and Black won't
(on average) be able to recover the *n* points he has paid.

However, there's a powerful counterargument to the above  I can put the
first black stone on the board as well as any professional can. And now,
assuming I am playing an equally weak human, it's White who suffers most
from the imperfection of our subsequent moves.

Nick

On 5 November 2015 at 11:39, Petr Baudis  wrote:

> On Thu, Nov 05, 2015 at 09:03:38AM +0200, Petri Pitkanen wrote:
> > 2015-11-05 0:04 GMT+02:00 Hideki Kato :
> >
> > > The correct komi value assuming both players are perfect.  Or, black
> > > utilize his advantage (maybe in an early stage) perfectly.  Actual
> > > players, even strong pros, are not perfect and cannot fully utilize
> > > their advantages.  As a conclusion, white is favored.
> >
> > Let alone we do not have even sufficient understanding of perfect play to
> > say what is correct komi in absolute sense. Nor it is it even meaningful
> > concept. Correct komi is a komi that produces about 50/50 result.
> Obviously
> > komi that will result in 50/50 for professionals will probably favour
> white
> > in your average weekend tournaments. Just like in chess 1st move
> advantage
> > is clearly less meanigfull for weaker players than top professionals.
>
> I find the notion above really counterintuitive, personally.
>
> Do you have any statistical evidence for this?  I.e. increasing portions
> of white wins in even games as the player rating decreases?
>
> Petr Baudis
> ___
> Computer-go mailing list
> Computer-go@computer-go.org
> http://computer-go.org/mailman/listinfo/computer-go
>



-- 
Nick Wedd  mapr...@gmail.com
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Petri Pitkanen]

I do doubt that there is sufficient data available on Go as it is not
popular enough. But lets face it 7 points guaranteed profit is way easier
to utilize than initiative.

For chess it clearly visible quotation from wikipedia
"database between players with similar Elo ratings, commissioned by GM András
Adorján , showed
that as the players' ratings went up, the percentage of draws increased,
the proportion of decisive games that White won increased, and White's
overall winning percentage increased.[15]
 For
example, taking the highest and lowest of Adorján's rating categories of
1669 games played by the highest-rated players (Elo ratings 2700 and
above), White scored 55.7% overall (W26.5 D58.4 L15.2), whereas of
34,924 games played by the lowest-rated players (Elo ratings below 2100),
White scored 53.1% overall (W37.0 D32.1 L30.8)."
A clear difference and even the lowest category is pretty strong. Around
Elo 1500 1st move probably means next to nothing

Why would it be any different in Go? I think 1st move advantage is far less
for weak players Go than in chess, because doing a Null move is far  easier


2015-11-05 13:39 GMT+02:00 Petr Baudis :

> On Thu, Nov 05, 2015 at 09:03:38AM +0200, Petri Pitkanen wrote:
> > 2015-11-05 0:04 GMT+02:00 Hideki Kato :
> >
> > > The correct komi value assuming both players are perfect.  Or, black
> > > utilize his advantage (maybe in an early stage) perfectly.  Actual
> > > players, even strong pros, are not perfect and cannot fully utilize
> > > their advantages.  As a conclusion, white is favored.
> >
> > Let alone we do not have even sufficient understanding of perfect play to
> > say what is correct komi in absolute sense. Nor it is it even meaningful
> > concept. Correct komi is a komi that produces about 50/50 result.
> Obviously
> > komi that will result in 50/50 for professionals will probably favour
> white
> > in your average weekend tournaments. Just like in chess 1st move
> advantage
> > is clearly less meanigfull for weaker players than top professionals.
>
> I find the notion above really counterintuitive, personally.
>
> Do you have any statistical evidence for this?  I.e. increasing portions
> of white wins in even games as the player rating decreases?
>
> Petr Baudis
> ___
> Computer-go mailing list
> 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

Re: [Computer-go] Komi 6.5/7.5

[Christoph Birk]

On Nov 5, 2015, at 4:44 AM, Nick Wedd  wrote:
> However, there's a powerful counterargument to the above  I can put the first 
> black stone on the board as well as any professional can. And now, assuming I 
> am playing an equally weak human, it's White who suffers most from the 
> imperfection of our subsequent moves.

But White already got the komi ….
Christoph


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Petr Baudis]

On Thu, Nov 05, 2015 at 09:03:38AM +0200, Petri Pitkanen wrote:
> 2015-11-05 0:04 GMT+02:00 Hideki Kato :
> 
> > The correct komi value assuming both players are perfect.  Or, black
> > utilize his advantage (maybe in an early stage) perfectly.  Actual
> > players, even strong pros, are not perfect and cannot fully utilize
> > their advantages.  As a conclusion, white is favored.
> 
> Let alone we do not have even sufficient understanding of perfect play to
> say what is correct komi in absolute sense. Nor it is it even meaningful
> concept. Correct komi is a komi that produces about 50/50 result. Obviously
> komi that will result in 50/50 for professionals will probably favour white
> in your average weekend tournaments. Just like in chess 1st move advantage
> is clearly less meanigfull for weaker players than top professionals.

I find the notion above really counterintuitive, personally.

Do you have any statistical evidence for this?  I.e. increasing portions
of white wins in even games as the player rating decreases?

Petr Baudis
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

[Computer-go] Feature training with "offset"

[Detlef Schmicker]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Hi,

I would like to train features like in

http://www.remi-coulom.fr/Amsterdam2007/

but using DCNN probabilities as an additional not trained gamma, which
is always present. Did anybody try using an additional not trained
gamma (not necessarily DCNN)? Is there a reference?


Any suggestion would be great :)

Detlef
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.22 (GNU/Linux)
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=b31b
-END PGP SIGNATURE-
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Stefan Kaitschick]

I agree with Robert. 7 is still a hot candidate for all board sizes.

Stefan
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Petri Pitkanen]

Wrong kind of information if the issue is komi(or white win rate) vs
streng. EGD seems to have something but with grabby interface. So to get
any meaninfull data one would have request

Nothing as useful as chess-results
http://www.chess-results.com/tnr184639.aspx?lan=1=2=1=821 exists
where a relatively simple webspider could collect the information. It has
been done for players over 2000 and it shows that value of tempo increases
with players skill. Really no reason to doubt that in go. But no easy way
of getting the data exist.

KGS information would be useful but would need to be collected by the site
operator.  Webspider could overload the system and no interface exist that
would be usefull for collecting the data





2015-11-05 18:26 GMT+02:00 Michael Alford :

> On 11/5/15 7:19 AM, Petr Baudis wrote:
>
> On Thu, Nov 05, 2015 at 02:42:20PM +0200, Petri Pitkanen wrote:
>>
>
> I do doubt that there is sufficient data available on Go as it is not
>>> popular enough.
>>>
>>
>>How much data is enough?
>>
>
> May I suggest Remi's excellent goratings.org for data?
>
> Michael
>
> --
>
> http://en.wikipedia.org/wiki/Pale_Blue_Dot
>
>
> ___
> Computer-go mailing list
> 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

[Computer-go] CoDIT'16-Malta: CFP & Call for Special Sessions Proposal

[Achraf Jabeur Telmoudi]

Dear colleagues
This an invitation to participate in the 3rd International Conference 
onControl, Decision and Information Technologies, to be held in Malta,on April 
6-8, 2016. 

Website: www.codit2016.com  
CFP ( http://codit2016.com/CFP-CoDIT2016.pdf )

 Paperspresented will be published on IEEE Xplore.

 Call forspecial Sessions: We are looking forscientists desiring to organize 
focused Special Sessions: http://codit2016.com/index.php/special-sessions 

 
Topics of interest include, but are not limited to, the following:
Track I. Control and Automation:
Track II. Decision and Optimization:
Track III. Information Technologies and Computer Science
*
Important Dates
Special Session proposal: November 7, 2015

Deadline for submission of papers: December 7, 2015

Notification: February 2, 2016

*
Best regards
Dr. Achraf J TELMOUDI
Communication Chair of CoDIT'16

*
CoDIT'16 : 2016 International Conference onControl, Decision and Information 
Technologies
Website: www.codit2016.com  
CFP ( http://codit2016.com/CFP-CoDIT2016.pdf )*

 ___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Justin Blank]

Of course. You always use win rate, never margin for that type of analysis.

On Nov 5, 2015, at 9:57 AM, Darren Cook  wrote:

>> Of course, that's anecdata...anyone is welcome to prove or disprove this
>> old claim by analyzing the stats on KGS, or Tygem or wherever else.
> 
> Don't forget the distortion due to people knowing the komi, and playing
> to win, rather than playing to maximize their score.
> 
> Darren
> ___
> Computer-go mailing list
> 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

Re: [Computer-go] Facebook Go AI

[Hiroshi Yamashita]

Hi,

I tried darkforest by using Aya.

darkforest is 0-3 against 2d bot. (AyaMC2) (0 wins, 3 losses)
darkforest is 1-4 against 1k bot. (AyaBot4)
darkfores1 is 1-3 against 1k bot. (AyaBot4)

It looks like darkforest is 1k or 2k.
It plays very quickly, and plays ko very well. But sometimes
it fails ladder. Maybe pure DCNN without MC search?
darkforest is ver 1.0, darkfores1 is ver 1.1. a bit latest.

Regards,
Hiroshi Yamashita

- Original Message - 
From: "Nick Wedd" 

To: 
Sent: Wednesday, November 04, 2015 4:32 AM
Subject: Re: [Computer-go] Facebook Go AI



I think this Facebook AI may be the program playing on KGS as darkforest
and darkfores1.

Nick


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Erik van der Werf]

We know the true values for some small boards that were solved, and what
some strong human players believed those values should be before they were
solved. I think that for all cases the humans where either correct, or
under-estimating. I don't remember any over-estimations.

Here are some cases where humans underestimated:

size human migos
2x11   4  6
3x75 21
4x61  8
4x74 28
5x62  4

For more results see: http://erikvanderwerf.tengen.nl/mxngo.html

Perhaps this can be considered an indication that weaker players tend to
benefit less from the first player advantage.

Best,
Erik


On Thu, Nov 5, 2015 at 3:35 PM, Justin Blank  wrote:

> I have repeatedly seen people assert that komi must be different for
> players of different skill levels, and have repeatedly questioned it, but I
> have never seen anyone try to substantiate the claim. People who believe it
> find it obvious, but I don't. There are two pieces of evidence that I can
> think of:
>
> 1) that I believe someone played near-random engines against each other,
> and the correct komi was different (I cannot find where that was done). But
> that's so far from even DDK play that it's pretty useless.
> 2) I believe the old OGS (DGS?) forums included an analysis of their games
> and what the correct komi was. I cannot confidently quote the results. If
> those are the old OGS forums, I don't know if they even exist anymore.
>
> The data from go servers are freely available. Does white have a greater
> advantage for weaker players? It doesn't seem so--anecdotally when players
> posted their KGS stats, they varied a bit, but didn't seem to have a bias
> for White.
>
> Of course, that's anecdata...anyone is welcome to prove or disprove this
> old claim by analyzing the stats on KGS, or Tygem or wherever else.
>
> On Thu, Nov 5, 2015 at 2:03 AM, Petri Pitkanen  > wrote:
>
>> Let alone we do not have even sufficient understanding of perfect play to
>> say what is correct komi in absolute sense. Nor it is it even meaningful
>> concept. Correct komi is a komi that produces about 50/50 result. Obviously
>> komi that will result in 50/50 for professionals will probably favour white
>> in your average weekend tournaments. Just like in chess 1st move advantage
>> is clearly less meanigfull for weaker players than top professionals.
>>
>> So setting komi is not theroretical but statistical issue
>>
>> 2015-11-05 0:04 GMT+02:00 Hideki Kato :
>>
>>> The correct komi value assuming both players are perfect.  Or, black
>>> utilize his advantage (maybe in an early stage) perfectly.  Actual
>>> players, even strong pros, are not perfect and cannot fully utilize
>>> their advantages.  As a conclusion, white is favored.
>>>
>>> Hideki
>>>
>>> Aja Huang: 

Re: [Computer-go] Komi 6.5/7.5

[Petr Baudis]

On Thu, Nov 05, 2015 at 02:42:20PM +0200, Petri Pitkanen wrote:
> I do doubt that there is sufficient data available on Go as it is not
> popular enough.

  How much data is enough?  The games from KGS alone ought to be in
millions, and even EGD must carry at least tens of thousands of serious
tournament games.  A trend ought to be visible there, if you want to
substantiate that argument.

> But lets face it 7 points guaranteed profit is way easier
> to utilize than initiative.

  But is it?  I'm not that strong myself (EGF 2k) but I'm fine with
initiative.  It depends on playing style as much as anything.  But
others in this thread made my argument better.

Petr Baudis
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Kahn Jonas]

Of course. You always use win rate, never margin for that type of analysis.


Problem is, this is not relevant unless we have games with different
komi.
The winrate depends as much on the width of the distribution as on the
median. Hence for weak players, it may go closer to 50%, even if the
value of the komi that would be right gets farther away. More or less
the only relevant information at fixed komoi would be if the winrate is
more than 50% above (or below) a certain level, and less than 50% below
(or above) it.

Jonas
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Komi 6.5/7.5

[Petri Pitkanen]

and in Go one move advantage need that your 1st pro-level mode works
together with your subsequent non-pro-moves

2015-11-05 14:55 GMT+02:00 Christoph Birk :

>
> On Nov 5, 2015, at 4:44 AM, Nick Wedd  wrote:
> > However, there's a powerful counterargument to the above  I can put the
> first black stone on the board as well as any professional can. And now,
> assuming I am playing an equally weak human, it's White who suffers most
> from the imperfection of our subsequent moves.
>
> But White already got the komi ….
> Christoph
>
>
> ___
> Computer-go mailing list
> 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