I think I''m starting to understand how Neural Networks work. :>) It seems like they would be useful for analyzing large amounts of data or finding the best way to do something. However, could you use a Neural Network for pathfinding, enemy units knowing when to attack/defend/retreat, or formulating team strategies? Or would some other paradigm be more effective for this? -Thanks http://abyssgames.cjb.net

if first you trained the ANN to its near-fullest potential, and then implemented it statically, with many optimizations, it could still be effective (cpu-usage wise). This of coarse depends on what kind of problem your trying to solve, obviously the NN could not evolve anymore, so learning is out the window ^^. However, for some problems (such as path finding) you could in fact train a neural net to do this using some clever input methods, but something like A* would be a far better (and faster) decision. .sen