0
$\begingroup$

In the late 1950s, Allen Newell and Herbert Simon have implemented a practical example for the general problem solver in the domain of computer chess. The project was called NSS chess (Newell, Simon, Shaw chess) and it was using heuristics, a goal generator and macro actions for playing chess. The advantage is, that symbolic AI related algorithmd need less computational effort and the chess engine will run on slower computers very well.

Unfortunately, the original NSS software isn't available in sourcecode and the documents from that time are gone. So my question is if the concept was cloned by modern programmers? The idea isn't using a brute-force solver but become dependent on heuristics, STRIPS like symbolic planning and goal stacks.

[1] Chess programming Wiki, NSS chess program, https://www.chessprogramming.org/NSS

$\endgroup$
0
$\begingroup$

I don't believe anyone has implemented the original NSS concept on a modern system but AlphaZero, developed by DeepMind (Google purchased them in 2014), is a great example of a heuristic based solution. AlphaZero is quite possibly the best AI chess player program, and also the best Go player. AlphaZero uses a Monte Carlo tree search algorithm along with a deep learning neural network.

You might find the following article of interest: How to build your own AlphaZero AI using Python and Keras.

Some related papers:

Slides from CS 760: Machine Learning, Spring 2018, Mark Craven and David Page: Reinforcement Learning with DNNs: AlphaGo to AlphaZero,

A couple of good videos:

From DeepMind's blog which also has a good video:

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.