Person Authentication Based on Biometric Traits Using Machine Learning Techniques
Gautam Kumar, Debbrota Paul Chowdhury, Sambit Bakshi, Pankaj Kumar · 2020
Internet of Things (IoT) is a concept of transferring data among interrelated physical devices, objects, or humans over a network without the interaction of humans or machine. Therefore, in IoT, one device collects and sends data while others receive and act on it. To communicate and to share information through interconnected devices, authentication of the sender and receiver is essential. When the goal is to improve security in IoT, traditional authentication techniques, such as knowledge-based authentication and token-based authentication, become a challenge. Therefore, researchers strongly recommend using biometrics whenever direct human access is required. However, a biometric system is also vulnerable to different types of attacks. Therefore, it is necessary to avoid and detect such attacks and secure IoT devices. Throughout this chapter, we discuss IoT and its applications, the types of security in IoT, the identification and verification process in biometrics, the vulnerability of components of biometric systems, methods to secure these components, and different types of attacks that can be made at different modules of a biometric system along with machine learning techniques to detect these attacks. We also investigate the various biometric traits used to authenticate end users using machine learning (ML) techniques, various ML algorithms, and methodologies for features extraction, matching, and classifications. Finally, a deep model is trained, and the performance of the model is evaluated on the Caltech face database and the UBIRIS.v1 iris dataset. The results of the deep model are compared with traditional ML techniques.