Comparative Study of Principal Component Analysis and Independent Component Analysis

Abstract

Face recognition is emerging as an active research area with numerous commercial and law enforcement applications. This paper presents comparative analysis of two most popular subspace projection techniques for face recognition. It compares Principal Component Analysis (PCA) and Independent Component Analysis (ICA), as implemented by the InfoMax algorithm. ORL face database is used for training and testing of the system. The results show that for the task of face recognition, ICA outperforms PCA in terms of recognition rate and subspace dimensionality.

High Hydrostatic Pressure and Co-Fermentation by Lactobacillus rhamnosus and Gluconacetobacter xylinus Improve Flavor of Yacon-Litchi-Longan Juice

“…To correlate the volatile compound content with the different treated YLL juice, PCA and PLS-DA were performed, and contents of the 95 volatile flavor compounds in all YLL juice samples were taken as analytical variables. PCA reduced the dimensionality within the data set and detected similarities and/or differences among juice samples [34]. Table S3 shows that PCA and PLS-DA models have four principal components and the fitting parameters are R 2 X = 0.966, Q 2 = 0.859; the PLS-DA model contains four principal components and the fitting parameters are R 2 X = 0.997, Q 2 = 0.987.…”