Can I read Fast Convergence in Incremental PCA on EtoBox?

Fast Convergence in Incremental PCA by Quarktop is a document available to read on EtoBox.

What is Fast Convergence in Incremental PCA about?

This document summarizes the fast convergence rates of two incremental PCA algorithms: Krasulina (1969) and Oja (1983). The algorithms update their estimate of the top eigenvector in an online fashion as new data points arrive. The paper analyzes the convergence rates of these algorithms and shows that their potential functions decrease at a rate of O(1/n), meaning the estimates converge to the true top eigenvector at a rate proportional to the number of samples seen. It also discusses practical considerati

Author

Quarktop

Language

EN