looking for chess engines that do it not in the usual way

[flok]

Hi,

Most (if not almost all) chess engines decide on their moves by generating some tree of moves and deciding what to choose using the famous algorithm by Claude Shannon.

Now what I'm looking for is chess engines that do it differently. E.g. using some pattern matching algorithm, using a neural net, etc.
What I found is:
http://neural-chess.netfirms.com/ - the distributed chess project
http://samba.org/KnightCap/ - uses temporal difference learning
not too much yet unfortunately

[Nid Hogge]

Symbolic is doing something like that. But it's not released.

[flok]

Symbolic is doing something like that. But it's not released.

Would you like to elaborate on that? Wat is Symbolic doing?

[hgm]

Now what I'm looking for is chess engines that do it differently.

You overlooked N.E.G. 0.3d. It has no search at all. It just picks a move from the set of legal moves by counting how often the toSquare is attacked and defended, (and by what), and what is on it. And if the move is a check or not.

[Jim Ablett]

Hi Folkert,

Have you come across P.ConNerS (Parallel Controlled Conspiracy Number Search) by Ulf Lorenz before ?

http://wwwcs.uni-paderborn.de/fachberei ... nners.html
http://tinyurl.com/yvbp83

To quote from the web page >

P.ConNerS uses a non-conventional, non-alphabeta search algorithm. The search algorithm tries not only to maximize the search depth, but also tries to guarantee that even when one leaf-value changes, the result stays the same. A conspiracy 2 search may be interpreted as a special, global arrangement of a lot of so called singular extensions. As a result, it domain-independently searches highly selective and irregular game trees.

Jim.

[Marc Lacrosse]

Hi Jim

Please do check your PM box.

Marc

[Dr.Wael Deeb]

Hi Jim

Please do check your PM box.

Marc

Hey,don't tell me that you're sending master Jim a source code to compile

[Mike S.]

But does it make sense? Are there any alternatives known, without search(*), which achieve something like 2000+ human Elo strenghts? Chess is a mixture of several elements, such as knowledge, style, intuition, and calculation. But without calculation for some moves ahead, a non-search program will be a victim of the simplest tactics all the time. I don't think this approach makes sense.

*) I am not referring to alternative searchers such as P.ConNerS, which was strong and has won a human Grandmaster tournament (it was the first GM tournament win of a computer).

[hgm]

Indeed, I don't think that Chess can be played well without search. Even Human players employ (highly selective) alpha-beta search for getting their move.

But that doesn't mean that alternative approaches are not useful. They can always be combined with search, by employing the algorithm to each end leave of a normal search, and employ it in internal nodes to select moves to be searched., searched first, or searched deeper.

[Tony Thomas]

What's the point of using alternate aproaches that result in an engine that is almost as weak as a random move generator?