Chess engines that use a neural network • page 1/1 • General Chess Discussion • lichess.org

The AI chess engines seem to be gaining some ground over the chess engines that do not have a neural network to learn from their mistakes. Humans play by memory. They do not use assistance like engine books. An engines memory needs to be stored somewhere like an opening book, but it should not be using assistance from games they never played.

The following list shaows lots of engines with ratings.
http://www.computerchess.org.uk/ccrl/404/

Can we discover which engiens have an artificial inteligence that does self-learning.
Have all the engines playing against other engines used their own games as opening books?

If the chess engine can not learn, then it should not be part of this list in this forum post.

If we search the internet we should be able to find all the AI self learning chess engines.

Here are some ... , Can you add to this list without repeating the links.

Neural networked chess engines:

Alphazero - Deepmind.com
Leela Chess Zero (LC0) - lczero.org
Fat fritz - https://en.chessbase.com/post/fat-fritz-what-on-earth-is-that

The AI chess engines seem to be gaining some ground over the chess engines that do not have a neural network to learn from their mistakes. Humans play by memory. They do not use assistance like engine books. An engines memory needs to be stored somewhere like an opening book, but it should not be using assistance from games they never played. The following list shaows lots of engines with ratings. http://www.computerchess.org.uk/ccrl/404/ Can we discover which engiens have an artificial inteligence that does self-learning. Have all the engines playing against other engines used their own games as opening books? If the chess engine can not learn, then it should not be part of this list in this forum post. If we search the internet we should be able to find all the AI self learning chess engines. Here are some ... , Can you add to this list without repeating the links. Neural networked chess engines: Alphazero - Deepmind.com Leela Chess Zero (LC0) - lczero.org Fat fritz - https://en.chessbase.com/post/fat-fritz-what-on-earth-is-that

drmrboss edited

#2

All neural network engines are copy cat of Leela and most of them are top 15 engines

Fat Fritz ( leela engine but different Neural Network, NN )
2.leelenstein ( leela engine but different NN )
Allie ( own engine + leelenstein NN )

These are not mentioned in ccrl

Stooves ( I think own engine+ different NN)
Scorpio ( own engine + Leela NN)
Antifish ( Leela engine + different NN)
Beta one ( Leela engine + different NN)
Darkqueen ( I think Leela engine+ different NN)

All neural network engines are copy cat of Leela and most of them are top 15 engines 1. Fat Fritz ( leela engine but different Neural Network, NN ) 2.leelenstein ( leela engine but different NN ) 3. Allie ( own engine + leelenstein NN ) These are not mentioned in ccrl 4. Stooves ( I think own engine+ different NN) 5. Scorpio ( own engine + Leela NN) 6. Antifish ( Leela engine + different NN) 7. Beta one ( Leela engine + different NN) 8. Darkqueen ( I think Leela engine+ different NN)

Toscani

#3

All languages originated somewhere. They all have something in common. Chess engines get there ideas from different search criterias. It's that mix of criterias that makes it original. With the network learning idea, the different chess engine builds now have a better chance in playing better chess. I think, any engine that cannot learn from their mistakes, is going to be part of the past.

Thanks for the list. I hope their are others that can add to the list of NN Chess Engines.
Even if they maybe based on Lc0 or other algorithms.

--------------------------

NN Chess engine list with their algorithms listed could become useful for coders.
They could then look at the list and try to create something that has not yet been compiled.

https://www.chessprogramming.org/Search#Search_Algorithms

Quiescence Search (QS)
Alpha-beta tree (AB)
Monte Carlo Tree Search (MCTS)
Minmax (MM)
Neural Network (NN)

It's the right combination of chess pseudocode that's going to make the difference.

All languages originated somewhere. They all have something in common. Chess engines get there ideas from different search criterias. It's that mix of criterias that makes it original. With the network learning idea, the different chess engine builds now have a better chance in playing better chess. I think, any engine that cannot learn from their mistakes, is going to be part of the past. Thanks for the list. I hope their are others that can add to the list of NN Chess Engines. Even if they maybe based on Lc0 or other algorithms. -------------------------- NN Chess engine list with their algorithms listed could become useful for coders. They could then look at the list and try to create something that has not yet been compiled. https://www.chessprogramming.org/Search#Search_Algorithms 1. Quiescence Search (QS) 2. Alpha-beta tree (AB) 3. Monte Carlo Tree Search (MCTS) 4. Minmax (MM) 5. Neural Network (NN) It's the right combination of chess pseudocode that's going to make the difference.

Nordlandia

#4

CrazyAra - an NN Crazyhouse engine.