A Generalized Maxentropic Inversion Procedure for Noisy Data
Science Direct Working Paper No S1574-0358(04)70549-3
17 Pages Posted: 26 Mar 2018
Date Written: July 2001
Abstract
In this note we present a way of solving linear inverse problems with convex constraints and noisy data. Combining the methods of maximum entropy in the mean and the generalized maximum entropy, we are able to treat both the noiseless and the noisy cases as conceptually the same problem. This new generalized inversion procedure is easy to apply and compute and is useful for a large class of models in the natural and social sciences. Three detailed examples are developed and discussed.
Keywords: Constrained Linear Inverse Problems, Maximum Entropy, Noisy Data, 34A55, 54C70, 62J12, 49N30
Suggested Citation: Suggested Citation