Feature Extraction of High-Dimensional Structures for Exploratory Analytics

Abstract

This report summarizes lessons learned in a study conducted at the U.S. Army Research Laboratory (ARL) for visual examination of high-dimensional data (HDD). The initial effort included feature extraction (FE), as opposed to feature selection, of HDD structures for display. FE to a two- or three-dimensional Euclidean space allows exploration of underlying structure of HDD, even though the meaning of the actual data is obscured (latent variables). Some discussion of the FEs considered is given; further details can be found at URLs provided. Application of visual analytics technology (interaction/navigation within visualization) allows additional knowledge discovery. Research continues at ARL for the development of a method to gain insight into HDD, particularly in the application of an analytic strategy to terrorist data.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 2013
Accession Number
ADA585756

Entities

People

  • Andrew M. Neiderer

Organizations

  • United States Army Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Data Visualization
  • Dimensionality Reduction
  • Extraction
  • Feature Extraction
  • Feature Selection
  • Fuzzy Sets
  • Information Science
  • Mathematics
  • Military Research
  • Set Theory
  • Three Dimensional
  • Two Dimensional
  • Visualizations

Readers

  • Database Systems and Applications
  • Geotechnical Engineering.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • Space