Convergence analysis of the NOJA algorithm using the ODE approach
Samir Attallah, Jonathan H. Manton, Karim Abed‐Meraim · Signal Processing · 2006
Some time ago, we proposed two normalized versions of Oja's algorithm (NOja and NOOja) and gave some results concerning their convergence performance for the estimation of the signal and noise subspaces of a sequence of random vectors. In this paper, we comment on NOja results using the ordinary differential equation (ODE) concept.