An evaluation of task performance capabilities in a network-centric system with neural networks usage

• Datum: 2025
• Hauptverfasser: Бойченко, А. В., Додонов, В. О., Залужний, В. Ф., Ізварін, Є. І.
• Format: Artikel
• Sprache: Ukrainisch
• Veröffentlicht: Інститут проблем реєстрації інформації НАН України 2025
• Schlagworte: neural network; capability; opportunity; artificial intelligence
• Online Zugang: https://drsp.ipri.kiev.ua/article/view/354610
• Назва журналу: Data Recording, Storage & Processing

The study considers the problem touching upon analyzing the military unit capabilities. The ability to accurately assess and predict the functional potential of the units is outstanding for effective strategic and operational planning. Established methodologies use strict calculation algorithm and mandatory formulas. Their limitation is an inflexibility during the crucial stages of result verification and validation. To overcome these constraints, a neural network–based framework for capability assessment is proposed. A key novelty is providing a continuous scale of assessment, offering a more nuanced view than traditional classifications. The formation of a training dataset can be carried out using two approaches. The first approach relies on human expert evaluation, where a specialist determines the capability level of a unit based on a predefined scale. The second approach involves the automated generation of labels using formulas, where a specific importance coefficient is assigned to each parameter. Within the suggested distributed architecture, each neural networks handle capability evaluation levels, ensuring modularity, scalability, and system resilience. To build the system can be used open software libraries for machine learning, data collection, and analysis. The approach simplifies the update process for individual system nodes. Input and output parameters can be changed or edited without the need to retrain the network. The results can be deployed into analytical support systems for military commanders and staff at all levels of command and control. The implementation of the system contributes directly to mission success and force preservation by reducing the probability of assigning unattainable or impossible tasks to subordinate units. Fig.: 5. Refs: 17 titles.