Enhanced Detection Algorithm for Navy Relevant Chemical Sensing

Abstract

This effort describes testing of alternate detection algorithms including evaluating the applicability of deep learning methods such as the application of Multivariate Long Short-Term Memory Fully Convolutional Network (MLSTM-FCN) approaches. The work is specifically intended to generate additional capabilities for these prototype devices allowing for their utilization in providing enhanced monitoring of chemical threats to the health of personnel in confined spaces both during a critical exposure event and as a result of long duration low level exposures. It was determined that deep learning methods are not appropriate for the type of data collected. The Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test was found to have equivalent accuracy to the already developed algorithm.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 15, 2021
Accession Number
AD1155763

Entities

People

  • Anthony P. Malanoski
  • Brandy J. Johnson
  • Dan Zabetakis
  • Jeff S. Erickson
  • Jerome E. Alvarez
  • Scott N Dean

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Autonomy
  • Sensors

DTIC Thesaurus Topics

  • Accuracy
  • Acquisition
  • Algorithms
  • Artificial Intelligence Software
  • Biological Sciences
  • Data Analysis
  • Deep Learning
  • Detection
  • Detectors
  • Engineering
  • Machine Learning
  • Materials
  • Military Research
  • Recognition
  • Risk Analysis
  • Standards
  • Test Methods

Readers

  • Neural Network Machine Learning.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Space
  • Space - Space Objects