Is there an intelligent variation of prinicipal component analysis

Well,i had a matrix of order 45*7. Each column represented a particular type of thermocouple. So i used principal component analysis to draw score plots and identify each of them using minimum distance rule.

My question was that i want to use intelligent techniques or some other intelligent algorithm to identify them. Is there an intelligent variation of principal component analysis that may be faster like ANN techniques or something which i can use to extend my work..

Answer 1

[a]

Yes, there are many nonlinear dimensionality reduction techniques, such as local linear embedding, that are generalizations of PCA. Check out

1. http://en.wikipedia.org/wiki/Nonlinear_dimensionality_reduction

2. http://www.amazon.com/Nonlinear-Dimensionality-Reduction-Information-Statistics/dp/0387393501/ref=sr_1_1?ie=UTF8&qid=1382502916&sr=8-1&keywords=nonlinear+dimensionality+reduction

3. http://homepage.tudelft.nl/19j49/Matlab_Toolbox_for_Dimensionality_Reduction.html