Hossein Azizpour

Hossein Azizpour is an Associate Professor in Machine Learning at KTH Royal Institute of Technology, affiliated with the Division of Robotics, Perception and Learning (RPL). He holds associated roles at the Science for Life Laboratory, Swedish e-Science Research Centre, WASP, and Digital Futures. His research focuses on deep learning under uncertainty, model interpretability, feature selection, and applications in life sciences, physics, and earth observation.

• Research Themes: (1) Deep learning under uncertainty (Bayesian methods, semi-supervised learning), (2) Interpretable AI (feature attribution, understanding neural networks), (3) Generative & foundation models (Vision Transformers, diffusion models), (4) Applications in medical imaging, fluid dynamics, and earth observation.
• Recent Work Trends: His publications emphasize vision transformers, uncertainty quantification, and cross-domain applications. Key areas include model optimization (e.g., Patchdropout), medical AI (e.g., breast cancer risk prediction), and physics-informed deep learning.
• Scientific Awards:
  • Best Paper Award, CVPR DeepVision Workshop (2015)
  • Google Travel Grant, CVPR DeepVision Workshop (2015)
  • Best Paper Runner-Up, CVPR DeepVision Workshop (2014)
  • Best Industry Paper Prize, BMVC (2013)
  • Outstanding Reviewer, CVPR (2013)
• Advising & Collaborations: Supervises PhD students in AI for physics applications (e.g., fluid dynamics), medical imaging, and computer vision. Collaborates with institutions like AstraZeneca, Alan Turing Institute, and Meta.
• Labs & Teams: Leads the Robotics, Perception and Learning (RPL) division’s research and co-organizes the Computer Vision Reading Group at KTH.