Determining the Optimizer Performance of CNN for Face Recognition

Abstract

Face recognition is the most widely used method of identifying and verifying an individual. Face recognition systems can detect and identify people in photos, videos, and in real time. Face recognition is widely used for biometric and security purposes because it can capture a wide range of information and does not require physical contact for verification. Deep learning is used to improve the efficiency of face recognition. A convolutional neural network (CNN) is a deep learning algorithm that has found widespread success in a variety of real-time applications. This algorithm overcomes the limitations of the machine learning algorithm. The CNN's performance is primarily determined by the network's architecture, optimizer, and parameters. There are several optimizers available to assist CNN obtain an optimal solution through quick convergence. The performance of several optimizers for different face datasets is tested in this work in order to discover the best optimizer for face recognition. This study gives a thorough grasp of several optimizers.