Artificial Intelligence in Biometric Systems
D Subitha, S G Rahul, Md Palash Uddin · 2024
Biometric systems are used for the identification and authentication of individuals using biological characteristics such as fingerprints, face, iris, palm print, gait, and voice. For example, fingerprint mapping, facial recognition, speech recognition, and retina scans are different forms of biometric technology that are used in various applications. The accuracy, adaptability, robustness, and speed of the traditional biometric systems can be enhanced using AI algorithms. The traditional biometric systems suffer from diverse environmental conditions, such as aging people, posture errors, etc. AI algorithms such as Convolutional Neural Network (CNN), Recurrent Neural Network, and Deep Neural Network (DNN) can help in developing biometric systems such as face and iris-based recognition that overcome all these difficulties. Simple classification algorithms such as KNN and Support Vector Machine (SVM) are the most common algorithms in fingerprint-based biometric systems. Further, more privacy can be provided using the Federated Learning Approach which is a decentralized process and Generative Adversarial Networks (GANs) are used to improve the robustness of biometric systems against adversarial attacks. The transfer learning approach is yet another technology used for systems in which the availability of data on a particular biometric feature is limited. The chapter gives an insight into the various AI algorithms that are used in biometric systems starting from the basic to the recent advances.