Character Recognition Analysis with Nonnegative Matrix Factorization

Abstract: In this paper, we analyze character recognition performance of three different nonnegative matrix factorization (NMF) algorithms. These are multiplicative update (MU) rule known as standard NMF, alternating least square (NMF-ALS) and projected gradient descent (NMF-PGD). They are most preferred approaches in the literature. There are lots of application areas for NMF such as robotics, bioinformatics, vision, sound and others. We use well known MNIST digit data set to test the performance of NMF, NMF-ALS and NMF-PGD. Experimental results show that NMF-ALS is the best and the worst one is NMF-PGD for these there algorithms in the meaning of accuracy. Therefore, we suggest NMF-ALS method can be used to analyze patterns on character recognition.