Topological principal component analysis for face encoding and recognition

Principal component analysis (PCA)-like methods make use of an estimation of the covariances between sample variables. This estimation does not take into account their topological relationships. This paper proposes how to use these relationships in order to estimate the covariances in a more robust way. The new method topological principal component analysis (TPCA) is tested using both face encoding and recognition experiments showing how the generalization capabilities of PCA are improved.