Title :
Against the convergence of the minor component analysis neurons
Author :
Taleb, Anisse ; Cirrincione, Giansalvo
Author_Institution :
LIS-INPG, Grenoble, France
fDate :
1/1/1999 12:00:00 AM
Abstract :
We investigate the dynamical behavior of some linear neurons devised for the minor component analysis. Contrary to their convergence theorems, our analysis reveals the existence of a divergence in a finite time
Keywords :
approximation theory; differential equations; eigenvalues and eigenfunctions; learning (artificial intelligence); neural nets; signal processing; stochastic processes; convergence; differential equation; divergence; eigenvalues; linear neurons; minor component analysis; neural nets; stochastic approximation; Artificial neural networks; Clustering algorithms; Convergence; Joints; Neural networks; Neurons; Pattern recognition; Power capacitors; Self organizing feature maps; Statistics;
Journal_Title :
Neural Networks, IEEE Transactions on
