(YIP) THEORY AND EFFICIENT ALGORITHMS FOR DYNAMIC AND ROBUST L1-NORM ANALYSIS OF TENSOR DATA

Abstract

The main objective of the proposed research is to develop theory and efficient algorithms for dynamic and robust analysis of multi-modal (tensor) data, based on L1-norm formulations. Specifically, we will first formulate Stochastic L1-norm Principal Component Analysis (S-L1- PCA) and investigate for the first time its theoretical underpinnings (graph of metric function, connection to batch L1-PCA, connection to standard stochastic PCA, etc.) Then, we will develop efficient online algorithms for the solution of S-L1-PCA, based on solid stochastic approximation theory. The developed methods will be accompanied by formal complexity analysis and stochastic convergence guarantees. Next, we will expand these methods for the analysis of tensor data, in the form of dynamic L1-Tucker decomposition. Specific considerations will be made for adaptation to various changes of the processed tensor. In addition, emphasis will be placed on the development of scalable algorithms that can be used in systems with limited computational resources. The developed algorithms will be rigorously tested on real data and various key applications.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 12, 2021
Source ID
FA95502010039

Entities

People

  • Panos P. Markopoulos

Organizations

  • Air Force Office of Scientific Research
  • Rochester Institute of Technology
  • United States Air Force

Tags

Fields of Study

  • Engineering

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Distributed Systems and Data Platform Development
  • Neural Network Machine Learning.