PresentationPDF Available

Extending Data Farming for Multi-scale Decision Support by Integrating Novel AI Technologies Multi-Dimensional Data Farming

Authors:
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 111/10/2023 | PAGE 1
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
Extending Data Farming for Multi-scale Decision Support by
Integrating Novel AI Technologies
Multi-Dimensional
Data Farming:
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 2
BLUF
Multi-Dimensional Data Farming (MDDF) automates simulation decision-
making by the integration of innovative Artificial Intelligence (AI) Techniques
and allows improved and fast decisions in highly complex multi-scale, multi-
domain, and multi-level hybrid war campaigns.
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 3
Data Farming
Codified by MSG-088 Data Farming in Support of NATO
The Realms of Data Farming (DF) collaboratively
connected by the DF Loop-of-Loops
Characterization or Optimization
NATO: Military Decision-Making
Improved Situational Awareness
Develop, analyze & refine COA
Informed & robust decisions
Proven and applied decision-support tool
Limitations for large multi-scale campaigns with many factions and operations interacting at various levels
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 4
Unconventional Campaign
Blue & Green armed forces vs. Red fighting in physical &
cyber domains (s. RSY MSG-197 in Bath, GBR, 2022)
Red launches a hybrid operation on Green, triggering a
Border Operation, in support of two strategic objectives:
force Green to divert resources away from the Campaign
lower Green public opinion in favor of the Campaign
Strategic level: multi-faction multi-domain combat model
(Attrition, Cyber, Epidemic ACE model)
Tactical level: multi-agent based simulations of the Border Operation (Map Aware Non-uniform Automata,
MANA)
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 5
Campaign Probability of Success
* alpha of supply for BLUE by GREEN
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 6
Border Operation
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 7
Embedding the Tactical Operation
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 8
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 9
Accelerating Automation with AI
Explainable AI
Allows a broader range of meta-models to be considered without sacrificing interpretability
Automated Machine Learning
Increases the level of automation in building meta-models
Bayesian Optimization with Gaussian Processes
Framework to find optima of black-box functions
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 10
DAVE (Design, Analyze & Visualize Experiments)
Design of Experiments: Intelligent sequential search through COA space using meta-model predictions
Analyze: Optimal balance of forces between Unconventional Campaign (UC) and Border Operation (BO)?
Visualize: Temporal playback of UC, spatio-temporal playback of BO to gain insight
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 11
DAVE Architecture & Implementation
Based on and using Data Farming Services
(DFS) from MSG-155
Dashboard + automized AI modules
Implemented standalone for
PoC/Demonstrator, to be integrated later into
service-oriented architecture of DFS
PoC status in 2023, extended to
Demonstrator for CWIX 2024
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 12
DAVE Demo
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 13
Summary
MDDF models for the first time the interplay of a long-time scale campaign and a short-time scale
operation, accounting for the complexity that arises in terms of time, sharing of resources and domains
involved.
MDDF shows how it is possible to combine AI techniques exploring operations at multiple scales (domain,
level, time) and optimize the probability of winning the campaign.
MDDF introduces a new multi-model approach to couple simulation models of different aggregations that
allow more comprehensive multi-level military planning and decision-making.
MDDF allows to accelerate the automation process making it efficient and effective in analyzing large and
complex multi-dimensional data spaces.
MDDF AI driven automation may extend the application of MDDF as a resource and time efficient
simulation support for wargaming in domains like mission support or training and education.
NORTH ATLANTIC TREATY ORGANIZATION
SCIENCE & TECHNOLOGY ORGANIZATION
NATO MODELLING AND SIMULATION GROUP
PUBLIC RELEASE 11/10/2023 | PAGE 14
Chairs MSG-186 Multi-Dimensional Data Farming
FINLAND
LtCol, Dr. Esa Lappi
Esa.Lappi@mil.fi
GERMANY
LtCol Stephan Seichter
stephanseichter@bundeswehr.org
CANADA
Prof. Dr. Uyen Bao
p.bao.u.nguyen@gmail.com
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.