Analysis of Complex-Valued Neural Network for Gender Recognition based on Face Image

Analysis of Complex-Valued Neural Network for Gender Recognition based on Face Image

Author		: SINDI AMILIA; MAHMUD DWI SULISTIYO; RETNO NOVI DAYAWATI
Published on	: ICoICT 2015

Abstract

Automatic gender recognition is an emerging problem in computer visions. An accurate gender recognition system can be used to reduce the search space in face recognition system for about half. However, since there is no definitive features of sexual dimorphism on human face that can be applied to all kind of face shapes from any race and age, there has not been any system that can differentiate human gender based on face image with 100% accuracy. Thus, in this experiment, we will use complex-valued neural network classification system in hope that it will enable the gender recognition system to reach 100% accuracy, or close to it. The methods proposed in this paper are as follows. After the face image processed by using local binary pattern to accentuate face texture and gradient filter to define face outline, the features of the face image is extracted by using histogram of oriented gradient. Then, the dimension of the resulted vectors is reduced by using principal component analysis. The final feature vectors is then used in neural network training and neural network testing processes. The results show that the average accuracy rate of real-valued neural network system is a bit lower than the average accuracy rate of complex-valued neural network, with 78.2% and 80.2% average accuracy rate, respectively. The results also show that complex-valued neural network system can achieve convergence rate four time faster than the real-valued neural network, where real-valued neural network system needs 12,7277 seconds to be convergent and complex-valued neural network system needs only 3,230.9 seconds to be convergent.