Explainable AI for Biometrics
Sarah Benziane, Kaouter Labed · 2024
Explainable AI has a cascading effect, causing increased transparency in decision-making processes, which fosters greater user trust and acceptance. This trust leads to wider adoption of AI technologies in sensitive applications like healthcare and finance, where understanding and accountability are essential. In this paper, we propose an explainable AI survey-based biometrics models. It consists first of a comprehensive review of current explainable AI techniques applied to biometrics, focusing on methods such as feature importance analysis, saliency maps, and rule-based models. We then assess these methods' effectiveness in enhancing transparency and interpretability within biometric systems. This paper proposes a comprehensive framework for integrating explainable AI techniques into biometric models to enhance transparency, trust, and usability. The framework leverages a combination of explainable AI techniques, including feature attribution methods, interpretable model architectures, and visualization tools, to enhance the transparency and accountability of biometric models.