Jump to content

  • Log In with Google      Sign In   
  • Create Account


Member Since 03 Mar 2009
Offline Last Active Sep 19 2012 07:00 AM

Posts I've Made

In Topic: Chess AI with Neural Networks

19 September 2012 - 07:00 AM

You could use the NN to build an evaluation function, as Alvaro suggested.

Basically, feed it things like 'passed pawns', 'player captures', 'opponent captures', etc.. All of the things you would calculate as part of a regular evaluation function become inputs in to the network. The network would output a score that determine how good (or bad) a particular position is.

In theory it should work just fine. In practice though it is going to be difficult to provide training data.

In Topic: Connect Four AI

24 June 2012 - 09:11 AM

Alvaro gives good advice.

Keep in mind that your code spend 99% of its time looking at useless positions. Anything you can do to get plausible positions explored early will eliminate a whole lot of work for the algorithm (thanks to alpha beta).

Iterative deepening and killer move heuristics can really speed things up substantially.

In Topic: State of the art game AI in 2012?

16 June 2012 - 09:49 AM

The state of the art is IBMs Watson. Nothing else comes close to it's achievements.

Second to that is Microsoft Kinect. It detects people and estimates poses in real time, even against cluttered backgrounds. Very artificially intelligent.

Most of the best AI is in things you don't notice... Path estimation to reduce the appearance of lag, algorithms to ensure realistic poses and animations, navigation algorithms, etc...

Most video game NPCs use scripted behavior or simple hard coded behaviors. You will find the most interesting AI where a machine needs to interface with reality, but at that point it tends to become very transparent.

Some UAVs have pretty slick AI for target tracking and line-of-sight planning. Googles self driving car is another example.

In Topic: Turn Based Strategy AI

08 June 2012 - 07:23 AM

Fantastic!! I'm so glad you ran in to the problem of determing a decent fitness function, and then solved it. Very educational. I think the important take away is that you found success when your fitness function was the sum of many different factors. It is critical that small tiny changes in fitness are easy and possible.

You illustrated perfectly why machine learning can be a cumbersome tool, but your results also demonstrated how fruitful the results can be. :). The next step, and I do hope you take it, would be to allow Henry to play against Henry. :)

Also interesting that you were able to use the GA to identify weaknesses in the game mechanics.

Well done, and thank you for sharing your progress!

In Topic: (Semi)automatic knowledge visualization (text to game/movie/cartoon/etc)

01 June 2012 - 06:10 AM

You might find this interesting. It's older, but has some elements of what you described.