General: 663

General: 664

Projects

Home



P2: Automated Parameter Selection Tool for Solution to Ill-Posed Problems

Author: Brianna Cash , Advisor: Dianne P. O'Leary (CS)


Problem Statement Presentation

Project Proposal

Abstract

In many ill-posed problems it can be assumed that the error in the data is dominated by noise which is independent identically normally distributed. Given this assumption the residual should also be normally distributed with similar mean and variance. This idea has been used to develop three statistical diagnostic tests to constrain the region of plausible solutions. This project aims to develop software that automates the generation of a range of plausible regularization parameters based on diagnostic tests.



MidYear Progress Report and Presentation

Final Presentation , Final Report