Inference-oriented model reduction- A new paradigm

Abstract

Inverse problems seek to characterize an underlying state or parameter of a system from indirectly observed data. Such problems are ubiquitous in science and engineering with applications including medical imaging, weather prediction, real-time flow estimation for flow control, and more. Solving inverse problems numerically is often extremely expensive because typical algorithms require the repeated evaluation of a high-dimensional computational model of the underlying physical system (one evaluation per sample or optimization iteration in typical inference algorithms). To reduce the cost of inference algorithms, efficient surrogate models for the underlying system are needed. The objective of this project is to develop a new paradigm for inference- oriented projection- based model reduction which exploits the known low-dimensional intrinsic dimensionality of inverse problems.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 06, 2025
Source ID
FA95502410105

Entities

People

  • Elizabeth Qian

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force

Tags

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Operations Research

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms