Artificial Neural Network Design and Parameter Optimization for Facial Expressions Recognition
This paper presents an Artificial Neural Network design and Neural Network parameter optimization for emotional recognition of classified facial expressions. The main goal in this paper is to teach computers to recognize three distinct human emotions from static images. Training and Testing dataset will be collected and a multilayer perceptron network will be built to implement an emotion classifier. Two excellent face databases are used to construct the training and testing datasets. Cross-validation techniques were used to compare the parameters of the Neural Network classifier and the types of activation functions.