This report demonstrates how machine learning (ML) can support assessment of military operations by describing and illustrating the use of ML in systematically extracting assessment-relevant insights from intelligence, operational, and media reporting. This approach can provide the commander with near-real-time insights from these data, often the best source of information on the efficacy of operations, that are objective and statistically relevant.

Leveraging Machine Learning for Operation Assessment

Daniel Egel, Ryan Andrew Brown, Linda Robinson, Mary Kate Adgie, Jasmin Léveillé, Luke J. Matthews

ResearchPublished May 9, 2022

The authors describe an approach for leveraging machine learning to support assessment of military operations. They demonstrate how machine learning can be used to rapidly and systematically extract assessment-relevant insights from unstructured text available in intelligence reporting, operational reporting, and traditional and social media. These data, already collected by operational-level headquarters, are often the best available source of information about the local population and enemy and partner forces but are rarely included in assessment because they are not structured in a way that is easily amenable to analysis. The machine learning approach described in this report helps overcome this challenge.

The approach described in this report, which the authors illustrate using the recently concluded campaign against the Lord's Resistance Army, enables assessment teams to provide commanders with near-real-time insights about a campaign that are objective and statistically relevant. This machine learning approach may be particularly beneficial in campaigns with limited or no assessment-specific data, common in campaigns with limited resources or in denied areas. This application of machine learning should be feasible for most assessment teams and can be implemented with publicly and freely available machine learning tools pre-authorized for use on U.S. Department of Defense systems.

Key Findings

Machine learning can be a powerful tool for supporting operation assessment

  • Data already collected by operational-level headquarters — intelligence reporting, operational reporting, and ambient data (social and traditional media) — are often the best available types of information about the enemy and partner forces and the local population. Yet they are rarely integrated into assessment, because they are often (1) not perceived as sufficiently objective, (2) not available in a structured format easily amenable to analysis, and (3) extremely numerous and require some effort to obtain and organize.
  • Machine learning (ML) tools, which can rapidly ingest and interpret large quantities of unstructured text, allow rapid, systematic, and objective analysis of these data, producing insights about the campaign that are objective and statistically relevant.
  • Supervised machine learning (SML) is the simplest approach for using ML to incorporate these data into the assessment process. In the SML approach, the assessment team first analyzes, by hand, a subset of the unstructured text and then applies ML algorithms to mimic the assessment team's analytical approach for the remaining data.
  • ML-derived data can provide a commander with near-real-time insights about a campaign, with each type of data (intelligence, operational, and ambient) providing a different lens for understanding a campaign's effects.
  • ML tools are particularly beneficial in campaigns with limited or no assessment-specific data — which is common in campaigns with limited resources or in denied areas.
  • This ML-based approach should be feasible for most assessment teams and can be implemented with freely available ML tools that are pre-authorized for use on U.S. Department of Defense classified systems.

Recommendations

  • Validate the supervised ML approach described in this report in a controlled exercise.
  • Explore how unsupervised ML can be used to inform operation assessment.
  • Implement modest standardization of operational reporting.
  • Improve archiving, discovery, and extraction of historical intelligence and operational reporting.
  • Expand assessment-specific discussions required in professional military education.

Order a Print Copy

Format
Paperback
Page count
96 pages
List Price
$23.00
Buy link
Add to Cart

Topics

Document Details

  • Copyright: RAND Corporation
  • Availability: Available
  • Year: 2022
  • Print Format: Paperback
  • Paperback Pages: 96
  • Paperback Price: $23.00
  • Paperback ISBN/EAN: 978-1-9774-0443-5
  • DOI: https://doi.org/10.7249/RR4196
  • Document Number: RR-4196-A

Citation

RAND Style Manual
Egel, Daniel, Ryan Andrew Brown, Linda Robinson, Mary Kate Adgie, Jasmin Léveillé, and Luke J. Matthews, Leveraging Machine Learning for Operation Assessment, RAND Corporation, RR-4196-A, 2022. As of September 23, 2024: https://www.rand.org/pubs/research_reports/RR4196.html
Chicago Manual of Style
Egel, Daniel, Ryan Andrew Brown, Linda Robinson, Mary Kate Adgie, Jasmin Léveillé, and Luke J. Matthews, Leveraging Machine Learning for Operation Assessment. Santa Monica, CA: RAND Corporation, 2022. https://www.rand.org/pubs/research_reports/RR4196.html. Also available in print form.
BibTeX RIS

Research conducted by

The research described in this report was sponsored by U.S. Army Special Operations Command and conducted by the Strategy, Doctrine, and Resources Program within the RAND Arroyo Center.

This publication is part of the RAND research report series. Research reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND research reports undergo rigorous peer review to ensure high standards for research quality and objectivity.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.

RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.