Welcome to IEEE SPS XAI 2025 Seasonal School !!!

About IEEE SPS XAI 2025

2025 IEEE Signal Processing Society (SPS) Cycle 2 Seasonal School on Explainable AI and Applications to Biometric Signal Processing is a one week short-term course organized by Computer Vision and Biometrics Lab (CVBL) at Indian Institute of Information Technology, Allahabad during 03-07 March 2025 under IEEE SPS UP Chapter. Explainable AI (XAI) refers to the set of techniques and methods in artificial intelligence that aim to make the outputs and decisions of AI systems understandable and interpretable by humans. This is particularly important in applications like biometric signal processing, where the decisions made by AI systems can have significant implications for individuals' security, privacy, and well-being. Feature Importance, Local Explanations, Model-agnostic Methods, and Visualizations are techniques used in XAI. Biometric signal processing involves analyzing physiological or behavioral characteristics of individuals for recognition or authentication purposes. This can include signals such as face, fingerprints, voice patterns, iris scans, and even biometric data derived from wearable devices. AI algorithms are increasingly being used to process and interpret these signals, but the opacity of these algorithms poses challenges in terms of trust, transparency, and accountability. XAI can address these challenges in terms of Transparency and Trustworthiness, Bias and Fairness, Compliance and Regulation, etc. XAI plays a crucial role in making biometric signal processing systems transparent, trustworthy, and compliant with regulatory standards. By providing explanations for AI decisions, XAI enhances user confidence and facilitates the responsible deployment of biometric technologies in various applications.

Welcome to the IEEE SPS Seasonal School on Explainable AI and its Applications to Biometric Signal Processing! This program offers a unique opportunity to explore the intersection of advanced AI techniques and the analysis of biometric data. Through expert-led discussions and interactions, the participants will gain valuable insights into how explainable AI can enhance the transparency and trustworthiness of biometric signal processing, paving the way for more accurate, interpretable, and ethical AI-driven solutions in fields like healthcare, security, and personal identification. This seasonal school aims to give the participants hands-on experience, right from the foundations, from some of the leading experts in the fields in the areas of Explainable AI (XAI), Biometric Signal Processing, advances in deep architectures and applications using different modality of data.

Objectives: This seasonal school aims to give the participants hands-on experience, right from the foundations, from some of the leading experts in the fields in the areas of Explainable AI, Deep Learning, Biometric Applications, and Signal Processing. The objectives of the seasonal school are highlighted as follows,,

• Building awareness towards Explainable AI approaches and applications in Biometric Signal Processing.
• Building technical capacity and communities of research students, educators, R&D and Industry persons in the trending areas of XAI, Biometric and Signal Processing.
• Providing tutorial and hands-on demo sessions, where the participants can experiment with concepts and methods.
• Facilitating the learning of participants through the internationally renowned experts in the domain.

IEEE SPS XAI 2025 Topics to be Covered:
The IEEE Seasonal School on Explainable AI and Applications to Biometric Signal Processing would involve covering various fundamental and advanced topics to provide participants with a comprehensive understanding. Following is the tentative topics:

• Introduction to Explainable AI and Biometric Signal Processing: Introduction to XAI, Definition and importance of explainability in AI, Motivation for using XAI in biometric signal processing, Biometric Signal Processing Basics
• Overview of biometric modalities (e.g., fingerprints, iris scans, voice recognition): Challenges and applications in biometric systems, Overview of Machine Learning and Deep Learning in Biometrics, Types of machine learning and deep learning models used in biometrics (e.g., SVMs, neural networks, CNNs, GANs, etc.), Training and evaluation metrics specific to biometric systems
• Techniques for Explainable AI: Interpretable Machine Learning Models, Linear models, decision trees, rule-based models, Pros and cons of different interpretable models, Local Interpretability Methods, Local interpretable model-agnostic explanations (LIME), SHAP (SHapley Additive exPlanations) values, Feature importance techniques
• Advanced XAI Techniques: Post-hoc Explanation Techniques, Layer-wise relevance propagation (LRP), Integrated gradients for sensitivity analysis, Counterfactual Explanations, Generating explanations for why a different decision might have been made, Ethical Considerations in XAI, Bias and fairness in biometric systems, Impact of XAI on privacy and data protection regulations
• Applications of XAI in Biometric Signal Processing: Face Recognition Systems, XAI techniques for explaining facial recognition decisions, Case studies and practical applications, other applications- Voice Recognition Systems, Health Monitoring and Wearable Biometrics, Real-time monitoring and diagnostic applications

IEEE SPS XAI 2025 Attractions:

• Keynotes, lectures and demo-sessions by world-renound international and national domain experts.
• Comprehensive coverage of XAI, deep learning and applications, and biometric solutions.
• Sponsored by IEEE Singal Processing Society (IEEE SPS).
• Minimum registration fee.
• Coffee break and Banquet access to registered and confirmed participants at IIIT Allahabad for the seasonal school.