Rate My Professor Behzad Bozorgtabar

Behzad Bozorgtabar is an Associate Professor of AI and Computer Vision at Aarhus University’s Department of Electrical and Computer Engineering, where he founded and heads the A3 Lab for Adaptive & Agentic AI. He is also a Guest Professor at the École Polytechnique Fédérale de Lausanne (EPFL). Bozorgtabar earned his PhD in Computer Vision from the University of Canberra in 2016, with a thesis titled "Vision-based Tracking for Sports Performance Analysis." His career includes serving as Head of the Computer Vision Team at EPFL’s Signal Processing Laboratory (LTS5), with joint affiliation to the Department of Radiology at Lausanne University Hospital (CHUV), and as a Postdoctoral Researcher at IBM Research Australia. He is a member of the European Laboratory for Learning and Intelligent Systems (ELLIS), affiliated with the Pioneer Centre for AI and Aarhus DIGIT Research Centre, and holds an Invited Professor position at EPFL’s School of Life Sciences.

Bozorgtabar’s research focuses on machine learning, computer vision, and medical image analysis, particularly self-supervised learning, test-time adaptation, multimodal foundation models, and agentic AI for trustworthy deployment in healthcare under domain shifts. He led the EPFL work-package in the EU Horizon 2020 ADAS&ME project involving over 30 partners. Notable roles include Senior Area Chair for NeurIPS 2026 and ICML 2026, Area Chair for CVPR and BMVC 2026, organizer of the Test-Time Updates Workshop at ICLR 2026, and the EPFL Computer Vision Talks series.

His influential publications feature "LATA: Laplacian-Assisted Transductive Adaptation for Conformal Uncertainty in Medical VLMs" (CVPR 2026, Oral), "ReservoirTTA: Prolonged Test-time Adaptation for Evolving and Recurring Domains" (NeurIPS 2025), "UniViT: Unifying Image and Video Understanding in One Vision Encoder" (NeurIPS 2025), "A Simple Framework For Open-Vocabulary Zero-Shot Segmentation" (ICLR 2025), and "CrIBo: Self-Supervised Learning via Cross-Image Object-Level Bootstrapping" (ICLR 2024, Spotlight). These works advance robust AI systems for real-world applications.