Life In 19x19

I would approach the bottom because it reduces white and expands black. But either is probably OK.

Marcel Grünauer wrote:

Uberdude wrote:Did anyone save a copy of the Nature paper? It's no longer available at https://storage.googleapis.com/deepmind ... ing-go.pdf

https://gogameguru.com/i/2016/03/deepmi ... ing-go.pdf

Nicer formatting: http://sci-hub.io/10.1038/nature16961

pasky wrote:That's not a compressed neural network but just a frequency-based specialized pattern classifier.

Ah, my mistake. I see the paper you referenced is specifically referring to compressing an existing neural network rather than training a smaller network.

So no, CAN'T do a "dumbed down version" for small machines. This is a very big neural net, lot of nodes, lots of "signals" going between nodes, takes a very powerful machine to get all that done within the allowed real time.


Are you absolutely certain about that? Since here somebody is arguing ...

Some thoughts about the continued benefit of AlphaGo for Go players.

DeepMind is a business concerned with Artificial Intelligence. AlphaGo is an interesting step in that direction, but it's not DeepMind's core business.

I don't see DeepMind in the business of making a commercial version of ...

It is not easy to write clear, consistent rules. Rule 6) could be more carefully stated. However, the writer or writers probably relied upon Rule 2) and did not worry about being punctilious.

2) Play: The players alternate in moving, with Black playing first. In handicap
games, White moves first ...

Oh, of course a player is allowed to recreate a previous board position by passing. That is how ever game ends

But I do think the AGA rules could be clarified here to say that Section 6 applies only to non-pass moves.

Black cannot pass because that would repeat a board position, so black must play elsewhere instead and eventually Black will run out of places to play and there will be no legal moves for him except filling his own eyes somewhere?

No, the AGA Superko rule allows repetition by a pass. With ...

AGA rules: those include a superko rule. So the cycle cannot be repeated without ko threats being played elsewhere. Games played by AGA rules are never a draw, they always have an outcome. If you are unsure about life and death, you should solve it by removing all dead groups (both players) by play ...

See m

Different rules handle Sending Two, Returning One in different ways. But in all of them there is no endless repetition when there is a single one of them on the board.

Thanks, that article clarifies things.

In this case,

$$Bc Is it alive?
$$ | . . . . . .
$$ | . O . . . .
$$ | O ...

Thanks. I can see what you mean by an error in transcription. I really just wanted to know if I had read these variations incorrectly somehow, since my goal is to improve my reading.

On Saturday I played my 300th Go game and to commemorate the occasion we somehow created a corner position that neither of use were sure the state of:

$$Bc Is it alive?
$$ | . . . . . .
$$ | . O . . . .
$$ | O . . . . .
$$ | X O O . O .
$$ | X X X O . .
$$ | . . X O . .
$$ | O . X O O ...

Ping. Anyone have any ideas here??

I'm reading Yuan Zhou's "Master Play: The Fighting Styles of Kato Masao and Seo Bong Soo" (which I'm really enjoying) and I came to a variation that I am not convinced of. Can you guys show me where I've gone wrong in my analysis? Here's the section from the book (From page 44):

Black tires to ...

For the math minded.

Given this game, { a || b | c }, a > b > c, to find the temperature, t, solve a - t = max(b, (b+c)/2 + t).

The mean value of the game is max(b, (2a+b+c)/4).

The temperature of the game is min(a-b, (2a-b-c)/4).

I find follow, could you explain your notation?
What does { a ...