English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper we consider the problem of learning variational models in thecontext of supervised learning via risk minimization. Our goal is to provide adeeper understanding of the two approaches of learning of variational modelsvia bilevel optimization and via algorithm unrolling. The former considers thevariational model as a lower level optimization problem below the riskminimization problem, while the latter replaces the lower level optimizationproblem by an algorithm that solves said problem approximately. Both approachesare used in practice, but unrolling is much simpler from a computational pointof view. To analyze and compare the two approaches, we consider a simple toymodel, and compute all risks and the respective estimators explicitly. We showthat unrolling can be better than the bilevel optimization approach, but alsothat the performance of unrolling can depend significantly on furtherparameters, sometimes in unexpected ways: While the stepsize of the unrolledalgorithm matters a lot (and learning the stepsize gives a significantimprovement), the number of unrolled iterations plays a minor role.

Learning Variational Models with Unrolling and Bilevel Optimization