ABSTRACT
Biometric authentication systems have widespread applications in national identification programs, banking, attendance system, and surveillance. Recent development and advances in technology has further led to the automation of these systems. Automated biometric authentication consists of several components, each of which are independent machine learning tasks. With the availability of large training data and access to sophisticated hardware, researchers have now begun focusing on the domain of deep learning to address the challenging covariates of biometrics. This chapter presents an introduction to deep learning by elaborating on its three main paradigms: restricted Boltzmann machine, autoencoder, and convolutional neural network. Each paradigm and some key architectures proposed in literature are discussed.
