Hi Chetan, I agree with everything you say. But..
AI is not anywhere near achieving human performance in how we do this evaluation, that's the whole problem. Humans do tree-searching for small numbers of moves ahead (and that number is vastly reduced by rejecting "bad moves" ie more intuitive evaluation), and then evaluate the choices to decide the move. The AI problem is not the searching, it's the evaluation. A single layer, single region CLA mathematically cannot do the evaluation unless it stores all board positions and sequences (combinatorially impossible), it takes a full hierarchy to abstract the heuristics you're talking about. Don't get me wrong, I am sure this is the only way for a machine to always win at chess! Regards, Fergal Byrne
_______________________________________________ nupic mailing list nupic@lists.numenta.org http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
