Title :
Equivalence of MSE and MaxEnt as objective function of PCA
Author :
Li, Hong-bao ; Hong-Bao Liu ; Ma, Ning-hua
Author_Institution :
Dept. of Educ. Adm., Hebei Univ., Baoding, China
Abstract :
Compare to MSE as PCA objective function to minimize, MaxEnt can at least achieve the same performance. In this paper, the authors proved that the two objective functions are equivalence in the sense that they achieve the optimal points in the same direction of principal components. Referring to the dimensionality reduction for the DCI benchmarking dataset, numerical experiments illustrate the equivalence of the two objective functions.
Keywords :
learning (artificial intelligence); maximum entropy methods; mean square error methods; principal component analysis; MSE; MaxEnt; PCA; UCI benchmarking dataset; dimensionality reduction; mean squared error minimisation; objective function; principal component analysis; Abstracts; Benchmark testing; Entropy; Face; Face recognition; KPCA; Maximum entropy; Objective function; PCA; Reconstruction error;
Conference_Titel :
Machine Learning and Cybernetics (ICMLC), 2012 International Conference on
Conference_Location :
Xian
Print_ISBN :
978-1-4673-1484-8
DOI :
10.1109/ICMLC.2012.6358903
