I was hoping, perhaps naively, that the neural network might somehow
'discover' the relevance of such things and the importance of them be
You are absolutely right with my relative understanding of NNW. The problem
is how fast can your NNW "discover". It might take longer than the age of
the universe. :) That's my limited understanding of the major limitation of
NNW for computer go. Feeding the network with a high level statistics will
be generally better, but I am not sure if a right play strategy is solely
dependant on those high level concepts without doing real calculation.