skip to content

The International Conference on Machine Learning (ICML) is a leading conference in machine learning, and this year Dr Angelica I Aviles-Rivero and Dr Jingwei Liang together with our head of our group Prof. Carola-Bibiane Schönlieb received an Outstanding Paper Award for the paper entitle “Tuning-free Plug-and-Play Proximal Algorithm for Inverse Imaging Problems”. This work was a joint collaboration with The Beijing Institute of Technology.

The Outstanding Paper awards are given to papers from the current year that are strong representatives of solid theoretical and empirical work in machine learning. For this edition the acceptance rate of the ICML was 21.8%, in which only two papers received such award (0.04%).

The awarded paper addressed a fundamental problem of Plug-and-Play algorithms: the need of manual parameter tweaking. For that, the authors proposed a tuning-free PnP proximal algorithm, which can automatically determine the internal parameters including the penalty parameter, the denoising strength and the terminal time. The core of the solution is a policy network for automatic search of parameters, which is learned via mixed model-free and model-based deep reinforcement learning. The carefully designed solution yield to state-of-the-art results.

Watch the Video Presentation:
Full Paper:
ICML Award Page: