A modified Mixture of FMLP Experts for face recognition
Nina Taheri Makhsoos, Alireza Hajiany, Reza Ebrahimpour, Ghodrat Sepidnam · 2008
In this paper we present a new face recognition model with principal component analysis in the feature extraction phase, and a Mixture of Fuzzy MLP Experts with Momentum term, in the recognition phase. We compared three different structures of neural network in which the average performance of Mixture of MLP Experts without Fuzzy MLP turned out to be 96% on a test set of 200 ORL face images. Our proposed model, using fuzzy MLPs as its expert networks, achieved a correct recognition rate of 98.9%. In fuzzy MLP, the ambiguity of each training sample is considered at the time of updating weights. Comparison with other algorithms demonstrate that our model performs better in terms of higher recognition rate, with smaller number of epochs in human face recognition.