Efficient Regularization of High-Dimensional Inverse Problems for Data Processing – EFFIREG

The need to efficiently process large amount of data has become ubiquitous in domains as wide as signal and image processing and machine learning. The objective of the EFFIREG project is to produce robust and computationally efficient data processing methods with theoretical performance guarantees. Using the mathematical tools provided by variational methods for inverse problems, overcoming typical difficulties associated to data processing amount to finding efficient regularization functions tuned to the problem in hand. Founded on recent advances in the study of regularization for low-dimensional models, this project aims at going beyond the heuristic approach to the design of regularization functions. Leveraging the expertise of its investigators, the EFFIREG project will in particular study inverse problems associated to applications found in medical image processing and machine learning.