Ensembles in face recognition: tackling the extremes of high dimensionality, temporality, and variance in data

Random subspaces are a popular ensemble construction technique that improves the accuracy of weak classifiers. It has been shown, in different domains, that random subspaces combined with weak classifiers such as decision trees and nearest neighbor classifiers can provide an improvement in accuracy. In this paper, we apply the random subspace methodology to the 2D face recognition task. The main goal of the paper is to see if the random subspace methodology can improve the performance of the face recognition system given the high dimensional data, temporal, and distribution variant data. We used two different datasets to evaluate the methodology. One dataset comprises of completely unique subjects for testing, and the other dataset comprises of the same subjects (both in training and testing) but images in the test set are captured at different times under different conditions.