Class-information-incorporated principal component analysis

Principal component analysis (PCA) is one of the most popular feature extraction methods in pattern recognition and can obtain a set of so-needed projection directions or vectors by maximizing the projected variance of a given training dataset in an unsupervised learning way, meaning that PCA does not sufficiently use the class label of given data in feature extraction. In this paper, a class-information-incorporated PCA (CIPCA) is presented with two objectives: one is to sufficiently utilize a given class label in feature extraction and the other is to still follow the same simple mathematical formulation as PCA. The experimental results on 13 benchmark datasets show its feasibility and effectiveness.