News for hackers. Stuff that matters.
Re:Introduction to Artificial Neural Networks – Part 1 by rdlecler1 on Saturday December 07, 2013 @04:10PM (#6867703)
It becomes a lot easier to think about these circuits if you realize that many of the connections are actually spurious. That is, if you remove them and systematically test the function over a range you'll realize that many of these weights have no information bearing value. For instance, if you have a w_i_j=0.01 this may be totally insignificant for any sigmoidal function and for all intents and purposes should be w_i_j=0 in which case this complex web of neural connections is trimmed, resulting in the underlying circuitry.

If you're interested I wrote a paper in Nature Systems Biology that shows this for artificial gene networks (just another branch of ANNs): Survival of the Sparsest: Robust Gene Networks are Parsimonious

There are no comments on this comment yet.