Algorithms on Flag Manifolds for Knowledge Discovery in N-way Arrays

Abstract

We proposed an approach for hyperspectral imagery classification that exploits the geometric framework of the Grassmannmanifold i.e., a parameterization of k dimensional subspaces of n-dimnsional space. The algorithm is particularly well suited to applications where sets of pixels are to be classified. Multiple pixels from a data class characterize the variability of the class information using a subspace representation. We use two metrics defined on the Grassmannian, chordal and geodesic, and one pseudometric, to compute pairwise distances between the points--subspaces. Once a distance matrix is generated, we use the classical multidimensional scaling to find a configuration of points with preserved or approximated original distances, thus realizing an embedding of the Grassmannian into Euclidean space. A sparse support vector machine (SSVM) trained in the embedding space simultaneously classifies embedded subspaces and selects a subset of optimal dimensions of the embedding for subsequentmodel reduction and data visualization.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 20, 2015
Accession Number
AD1000735

Entities

People

  • Michael Kirby

Organizations

  • Colorado State University

Tags

Communities of Interest

  • Biomedical

DTIC Thesaurus Topics

  • Compressed Sensing
  • Computational Science
  • Computer Vision
  • Data Analysis
  • Detection
  • Dimensionality Reduction
  • Geometry
  • Health Services
  • Hyperspectral Imagery
  • Information Science
  • Linear Algebra
  • Mathematical Models
  • Medical Personnel
  • Pattern Recognition
  • Signal Processing
  • Supervised Machine Learning
  • Warning Systems

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Graph Algorithms and Convex Optimization.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space