The performance of multi-layer neural network on face recognition system

Biometrics and Pattern Recognition have various applications that are found and brought into real-time application use. Face recognition consist mainly of three stages namely: Pre-processing, Feature Extraction and Classification. Neural Networks basically deals with adaptation, classification and rendering noisy values to optimal solution. In this work we illustrate performance and accuracy of the above approaches. Subspace is a plane embedded in a higher dimensional vector space, PCA is a standout amongst the best systems that have been utilized in image recognition and compression while KPCA is utilized in ascertaining PCA conversion in a mapping space by a nonlinear mapping function. FFNN is used for pattern recognition, FNN frequently have at least one hidden layers of sigmoid neurons followed by a yield layer of linear neurons. Multiple layers of neurons with nonlinear transfer function permits the system to learn connections amongst information and yield vectors. LVQ learn to characterize input vectors into target classes picked by the user.