Foundations of deep learning for scalable computational imaging

The Biomedical and Astronomical Signal Processing (BASP) laboratory at Heriot-Watt University Edinburgh (HWU), headed by Professor Yves Wiaux is recruiting a PhD student for research on foundational deep learning methodology for computational imaging in astronomy and/or medicine.

The BASP laboratory is developing cutting-edge research on all aspects of computational imaging, from theory and algorithms to a myriad of applications. Dr Wiaux is a Professor in the School of Engineering and Physical Sciences at HWU. He is also Honorary fellow at the School of Informatics of the University of Edinburgh (UoE), and Academic Guest at the Signal Processing Laboratories of the Ecole Polytechnique Fédérale de Lausanne (EPFL). The position is open in the context of large research projects funded by the UK Research Councils, aiming to develop a new generation of image reconstruction and analysis algorithms at the interface of deep learning, optimisation, and Bayesian sampling. The algorithms are intended to deliver simultaneously high resolution and dynamic range imaging, to include calibration and uncertainty quantification functionalities, and to enable scalability to unprecedented image sizes and data volumes. They will be designed, implemented, and validated for various applications in astronomy and medicine.

The research initiative, led by Prof. Wiaux, is supported by a unique team of international partners, among which the Digital Vision Centre of Université Paris-Saclay (CVN, Prof. JC. Pesquet), and the Institute of Computer Graphics and Vision of Graz University of Technology (ICG, Prof. Pock).

The PhD project will focus on foundational algorithmic developments, including: (i) the development of image reconstruction and analysis algorithms for large-scale imaging inverse problems, including estimation and uncertainty quantification, at the interface of deep learning and optimisation; (ii) the study of their theoretical properties and convergence guarantees in the context of deep learning and optimisation theories, including developing novel stochastic optimisation algorithms for the training of deep networks.

We are looking for outstanding candidates with a first Class Master’s Degree (or equivalent) in electrical engineering, applied mathematics, physics, computer science, or a related discipline. The student will be fully integrated into BASP under the supervision of Prof. Wiaux, with possible co-supervision by the relevant collaborators from the participating institutions.

The scholarship is fully-funded for 3.5 years, and open to international students. The start date is flexible from September 2025. Expressions of interest and full applications (single PDF including a motivation letter, full transcript of record, CV, and names of 2 references) should be sent by email to Prof. Wiaux ([email protected]). Please apply at your earliest convenience. Applications screening will start in January 2025 and continue until the position is filled.