Saved in:
Bibliographic Details
Main Authors: King-Roskamp, Matthew, Choksi, Rustum, Hoheisel, Tim
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.17916
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916541313843200
author King-Roskamp, Matthew
Choksi, Rustum
Hoheisel, Tim
author_facet King-Roskamp, Matthew
Choksi, Rustum
Hoheisel, Tim
contents We establish the theoretical framework for implementing the maximumn entropy on the mean (MEM) method for linear inverse problems in the setting of approximate (data-driven) priors. We prove a.s. convergence for empirical means and further develop general estimates for the difference between the MEM solutions with different priors $μ$ and $ν$ based upon the epigraphical distance between their respective log-moment generating functions. These estimates allow us to establish a rate of convergence in expectation for empirical means. We illustrate our results with denoising on MNIST and Fashion-MNIST data sets.
format Preprint
id arxiv_https___arxiv_org_abs_2412_17916
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Data-Driven Priors in the Maximum Entropy on the Mean Method for Linear Inverse Problems
King-Roskamp, Matthew
Choksi, Rustum
Hoheisel, Tim
Machine Learning
Optimization and Control
49N15, 49N30, 90C30, 68V10
We establish the theoretical framework for implementing the maximumn entropy on the mean (MEM) method for linear inverse problems in the setting of approximate (data-driven) priors. We prove a.s. convergence for empirical means and further develop general estimates for the difference between the MEM solutions with different priors $μ$ and $ν$ based upon the epigraphical distance between their respective log-moment generating functions. These estimates allow us to establish a rate of convergence in expectation for empirical means. We illustrate our results with denoising on MNIST and Fashion-MNIST data sets.
title Data-Driven Priors in the Maximum Entropy on the Mean Method for Linear Inverse Problems
topic Machine Learning
Optimization and Control
49N15, 49N30, 90C30, 68V10
url https://arxiv.org/abs/2412.17916