Re: CGOS source on github

--===============6772950181634131676==
Content-Type: multipart/alternative; boundary="0000000000008d8ec805b97df9d3"

--0000000000008d8ec805b97df9d3
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Claude - no, generally feeding liberty counts to neural networks doesn't
help as much as one would hope with ladders and sekis and large capturing
races.

The thing that is hard about ladders has nothing to do with liberties - a
trained net is perfectly capable of recognizing the atari, this is
extremely easy. The hard part is predicting if the ladder will work without
playing it out, because whether it works depends extremely sensitively on
the exact position of stones all the way on the other side of the board. A
net that fails to predict this well might prematurely reject a working
ladder (which is very hard for the search to correct), or be highly
overoptimistic about a nonworking ladder (which takes the search thousands
of playouts to correct in every single branch of the tree that it happens
in).

For large sekis and capturing races, liberties usually don't help as much
as you would think. This is because approach liberties, ko liberties, big
eye liberties, shared liberties versus unshared liberties, throwin
possibilities all affect the "effective" liberty count significantly. Also
very commonly you have bamboo joints, simple diagonal or hanging
connections and other shapes where the whole group is not physically
connected, also making the raw liberty count not so useful. The neural net
still ultimately has to scan over the entire group anyways, computing these
things.

On Fri, Jan 22, 2021 at 8:31 AM Claude Brisson via Computer-go <
[email protected]> wrote:

> Hi. Maybe it's a newbie question, but since the ladders are part of the
> well defined topology of the goban (as well as the number of current
> liberties of each chain of stone), can't feeding those values to the
> networks (from the very start of the self teaching course) help with larg=
e
> shichos and sekis?
>
> Regards,
>
>   Claude
> On 21-01-22 13 h 59, R=C3=A9mi Coulom wrote:
>
> Hi David,
>
> You are right that non-determinism and bot blind spots are a source of
> problems with Elo ratings. I add randomness to the openings, but it is
> still difficult to avoid repeating some patterns. I have just noticed tha=
t
> the two wins of CrazyStone-81-15po against LZ_286_e6e2_p400 were caused b=
y
> very similar ladders in the opening:
> http://www.yss-aya.com/cgos/viewer.cgi?19x19/SGF/2021/01/21/733333.sgf
> http://www.yss-aya.com/cgos/viewer.cgi?19x19/SGF/2021/01/21/733301.sgf
> Such a huge blind spot in such a strong engine is likely to cause rating
> compression.
>
> R=C3=A9mi
>
> _______________________________________________
> Computer-go mailing [email protected]://computer-go.org=
/mailman/listinfo/computer-go
>
> _______________________________________________
> Computer-go mailing list
> [email protected]
> http://computer-go.org/mailman/listinfo/computer-go
>

--0000000000008d8ec805b97df9d3
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi Claude - no, generally feeding liberty counts to neural=
 networks doesn&#39;t help as much as one would hope with ladders and sekis=
 and large capturing races.<div><br></div><div>The thing that is hard about=
 ladders has nothing to do with liberties - a trained net is perfectly capa=
ble=C2=A0of recognizing the atari, this is extremely easy. The hard part is=
 predicting if the ladder will work without playing it out, because whether=
 it works depends extremely sensitively on the exact position of stones all=
 the way on the other side of the board. A net that fails to predict this w=
ell might prematurely reject a working ladder (which is very hard for the s=
earch to correct), or be highly overoptimistic about a nonworking ladder (w=
hich takes the search thousands of playouts to correct in every single bran=
ch of the tree that it happens in).</div><div><br></div><div>For large seki=
s and capturing races, liberties usually don&#39;t help as much as you woul=
d think. This is because approach liberties, ko liberties, big eye libertie=
s, shared liberties versus unshared liberties, throwin possibilities all af=
fect the &quot;effective&quot; liberty count significantly. Also very commo=
nly you have bamboo joints, simple diagonal or hanging connections and othe=
r shapes where the whole group is not physically connected, also making the=
 raw liberty count not so useful. The neural net still ultimately has to sc=
an over the entire group anyways, computing these things.</div></div><br><d=
iv class=3D"gmail_quote"><div dir=3D"ltr" class=3D"gmail_attr">On Fri, Jan =
22, 2021 at 8:31 AM Claude Brisson via Computer-go &lt;<a href=3D"mailto:co=
[email protected]">[email protected]</a>&gt; wrote:<br></=
div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;bor=
der-left:1px solid rgb(204,204,204);padding-left:1ex">
 =20
   =20
 =20
  <div>
    <p>Hi. Maybe it&#39;s a newbie question, but since the ladders are part
      of the well defined topology of the goban (as well as the number
      of current liberties of each chain of stone), can&#39;t feeding those
      values to the networks (from the very start of the self teaching
      course) help with large shichos and sekis?</p>
    <p>Regards,</p>
    <p>=C2=A0 Claude<br>
    </p>
    <div>On 21-01-22 13 h 59, R=C3=A9mi Coulom wrote:<br>
    </div>
    <blockquote type=3D"cite">
     =20
      <div dir=3D"ltr">
        <div>Hi David,</div>
        <div><br>
        </div>
        <div>You are right that non-determinism and bot blind spots are
          a source of problems with Elo ratings. I add randomness to the
          openings, but it is still difficult to avoid repeating some
          patterns. I have just noticed that the two wins of
          CrazyStone-81-15po against LZ_286_e6e2_p400 were caused by
          very similar ladders in the opening:</div>
        <div><a href=3D"http://www.yss-aya.com/cgos/viewer.cgi?19x19/SGF/20=
21/01/21/733333.sgf" target=3D"_blank">http://www.yss-aya.com/cgos/viewer.c=
gi?19x19/SGF/2021/01/21/733333.sgf</a></div>
        <div><a href=3D"http://www.yss-aya.com/cgos/viewer.cgi?19x19/SGF/20=
21/01/21/733301.sgf" target=3D"_blank">http://www.yss-aya.com/cgos/viewer.c=
gi?19x19/SGF/2021/01/21/733301.sgf</a></div>
        <div>Such a huge blind spot in such a strong engine is likely to
          cause rating compression.</div>
        <div><br>
        </div>
        <div>R=C3=A9mi<br>
        </div>
      </div>
      <br>
      <fieldset></fieldset>
      <pre>_______________________________________________
Computer-go mailing list
<a href=3D"mailto:[email protected]" target=3D"_blank">Computer-g=
[email protected]</a>
<a href=3D"http://computer-go.org/mailman/listinfo/computer-go" target=3D"_=
blank">http://computer-go.org/mailman/listinfo/computer-go</a>
</pre>
    </blockquote>
  </div>

_______________________________________________<br>
Computer-go mailing list<br>
<a href=3D"mailto:[email protected]" target=3D"_blank">Computer-g=
[email protected]</a><br>
<a href=3D"http://computer-go.org/mailman/listinfo/computer-go" rel=3D"nore=
ferrer" target=3D"_blank">http://computer-go.org/mailman/listinfo/computer-=
go</a><br>
</blockquote></div>

--0000000000008d8ec805b97df9d3--

--===============6772950181634131676==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Computer-go mailing list
[email protected]
http://computer-go.org/mailman/listinfo/computer-go

--===============6772950181634131676==--