Research

Topics

Learnable Optimization AlgorithmsOptimization algorithms as statistical objects, with provable generalization and convergence guarantees Bilevel Optimization and Automatic DifferentiationDifferentiating through optimization algorithms and solution maps Inertial / Accelerated AlgorithmsHeavy-ball, iPiano and accelerated methods for non-smooth non-convex problems Dynamical System View on OptimizationContinuous-time dynamics as a lens on optimization algorithms Unification of Non-smooth Optimization AlgorithmsModel functions and Bregman distances as a common framework

Funded projects

Events

Earlier research

• Image and Video Segmentation – Motion-based video segmentation with point trajectories and variational methods (2011–2016)