A learning-enhanced projection method for solving convex feasibility problems
Author:
- Janosch RiegerJRJanosch Rieger
Preprint
Abstract
We propose a generalization of the method of cyclic projections, which uses the lengths of projection steps carried out in the past to learn about the geometry of the problem and decides on this basis which projections to carry out in the future. We prove the convergence of this algorithm and illustrate its behavior in a first numerical study.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- Union Averaged Operators with Applications to Proximal Algorithms for Min-Convex FunctionsJournal of Optimization Theory and Applications, 2018
- AIR Tools II: algebraic iterative reconstruction methods, improved implementationNumerical Algorithms, 2017
- Convergence analysis for column-action methods in image reconstructionNumerical Algorithms, 2016
- The convergence of the block cyclic projection with an overrelaxation parameter for compressed sensing based tomographyJournal of Computational and Applied Mathematics, 2015
- A Randomized Kaczmarz Algorithm with Exponential ConvergenceJournal of Fourier Analysis and Applications, 2008
- Convergence studies on iterative algorithms for image reconstructionIEEE Transactions on Medical Imaging, 2003
- Accelerating the convergence of the method of alternating projectionsTransactions of the American Mathematical Society, 2003
- Acceleration schemes for the method of alternating projectionsJournal of Computational and Applied Mathematics, 1989
- On the von Neumann alternating algorithm in Hilbert spaceJournal of Mathematical Analysis and Applications, 1986
- Row-Action Methods for Huge and Sparse Systems and Their ApplicationsSIAM Review, 1981