in signal processing, wavelet is very famous. whereas in bp neural network, they are the same principle. it is the infinite series. the only difference is that bp network is focus on auto-learning whereas warelet is for time domain convert to frequence domain.
in previous article, i have introdue several fuzzy theory and its application. but why neural network is belong to soft computing? in soft computing, the model is more flexible and robust than the normal one. vision of fuzzy theory is to sketch a fuzzy paradigm. but in neural network, is to avoid to sketch improcise directly. it is as same as GA algorithm.
before few days, i read my teammate's thesis, i found that to representation of computing we have a lot of method, such as formular, rules, and algorithm for procedure etc.
so, in capable of bp neural network, it can represent nonlinear function. and I can call it like this sentence: a bp neural network is one of representation of computing which can expresses nonlinear function with imprecise using infinite series.