Speaker: Dr. Moloud Abdar

Topic: Uncertainty Quantification for Deep Learning (UQ4DL)

Abstract: Deep learning models have achieved remarkable success across many domains, from computer vision and natural language processing to healthcare and social sciences. However, their deterministic nature often limits their reliability in high-stakes decision-making, where understanding what a model does _not_ know is as important as what it predicts. This talk introduces the principles and recent advances in uncertainty quantification (UQ) for deep learning. I will discuss key concepts of epistemic and aleatoric uncertainty, methods such as Bayesian deep learning, ensemble techniques, and Variational Inference, and how they enhance model interpretability, robustness, and trustworthiness. Practical applications will be presented across domains including medical diagnosis, vision-language models (VLMs), and remote sensing. The talk will conclude with open challenges and future research directions for integrating uncertainty-aware AI into real-world, ethical, and transparent decision-making systems.

Biography: Moloud Abdar received the Ph.D. degree in Computer Engineering/Science from Deakin University, Australia, in 2022. He is currently a Senior Data Scientist and AI Lead at The University of Queensland, Australia. His research interests include trustworthy and explainable artificial intelligence, machine learning, uncertainty quantification, and computational modelling for high-stakes decision-making. Dr. Abdar has published over 90 peer-reviewed papers in leading journals and conferences, including _IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)_, _Transactions on Machine Learning Research (TMLR)_, _Information Fusion_, _IEEE Transactions on Industrial Informatics (TII)_, _IEEE Transactions on Evolutionary Computation (TEVC)_, CVPR, ICCV, and BMVC. His work has received more than 10,000 citations with an H-index of 42, ranking him among the Top 2% of Scientists worldwide (Stanford University, 2022-2025) and the Top 0.5% of Global Scholars (ScholarGPS, 2024-2025). His research excellence has been recognized through multiple international honors, including the 2026 IEEE Transactions on Artificial Intelligence Outstanding Paper Award (IEEE Computational Intelligence Society), the 2025 Rising Star of Science Award (Research.com), the 2024 Australian Pattern Recognition Society Early Career Researcher Award, and the 2022 Best Survey Award (_Information Fusion_, Elsevier). He was also named Outstanding Early Career Researcher at Deakin University’s Institute for Intelligent Systems Research and Innovation (2022). Dr. Abdar leads and collaborates on international projects with institutions such as Oxford, Cambridge, Harvard, National University of Singapore, Nanyang Technological University, and The University of Melbourne, as well as with industry partners including Google, NVIDIA, Meta, Amazon, and Qualcomm AI Research. His research bridges theoretical AI foundations with applied computational methods, focusing on ethical, interpretable, and uncertainty-aware decision systems.