Bilevel Optimization with Nonsmooth Lower Level Problems
International Conference on Scale Space and Variational Methods in Computer Vision (SSVM), Springer, LNCS, Vol.9087: 654--665, 2015
Abstract: We consider a bilevel optimization approach for parameter learning in nonsmooth variational models. Existing approaches solve this problem by applying implicit differentiation to a sufficiently smooth approximation of the nondifferentiable lower level problem. We propose an alternative method based on differentiating the iterations of a nonlinear primal--dual algorithm. Our method computes exact (sub)gradients and can be applied also in the nonsmooth setting. We show preliminary results for the case of multi-label image segmentation.
Publisher's link
Images and movies
BibTex reference
@InProceedings{OB15a, author = "P. Ochs and R. Ranftl and T. Brox and T. Pock", title = "Bilevel Optimization with Nonsmooth Lower Level Problems", booktitle = "International Conference on Scale Space and Variational Methods in Computer Vision (SSVM)", series = "Lecture Notes in Computer Science", volume = "9087", pages = "654--665", month = " ", year = "2015", editor = "J.-F. Aujol, M. Nikolova, N. Papadakis", publisher = "Springer", note = "Awarded the SSVM 2015 Best Paper Award", url = "http://lmb.informatik.uni-freiburg.de/Publications/2015/OB15a" }