Workshop on Recent Advances in

Learned Regularisation

20-21 February 2025
66-72 Gower Street, UCL

Workshop Themes

Deep learning approaches have been one of the predominant methodological paradigms for inverse problems over the last decade. Of particular note are learned regularisation approaches, which allow a synergy between data-driven approaches and theoretical framework from classical theory. This is a growing research area that includes unrolled optimisation, learning-to-optimise, regularisation by denoising, adverserial regularisation, bilevel learning, and other learning frameworks.

The goal of this workshop is to bring together researchers working at the cutting edge of learned regularisation and their applications.

Organisers