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

Numerous tasks at the core of statistics, learning and vision areas arespecific cases of ill-posed inverse problems. Recently, learning-based (e.g.,deep) iterative methods have been empirically shown to be useful for theseproblems. Nevertheless, integrating learnable structures into iterations isstill a laborious process, which can only be guided by intuitions or empiricalinsights. Moreover, there is a lack of rigorous analysis about the convergencebehaviors of these reimplemented iterations, and thus the significance of suchmethods is a little bit vague. This paper moves beyond these limits andproposes Flexible Iterative Modularization Algorithm (FIMA), a generic andprovable paradigm for nonconvex inverse problems. Our theoretical analysisreveals that FIMA allows us to generate globally convergent trajectories forlearning-based iterative methods. Meanwhile, the devised scheduling policies onflexible modules should also be beneficial for classical numerical methods inthe nonconvex scenario. Extensive experiments on real applications verify thesuperiority of FIMA.

On the Convergence of Learning-based Iterative Methods for Nonconvex Inverse Problems