MOSCOW, October 12. /TASS/. The introduction of a new system for training military-grade neural networks will, in the long term, make the Russian Armed Forces more autonomous, Alexander Stepanov, a military expert at the Institute of Law and National Security of the Russian Presidential Academy of National Economy and Public Administration (RANEPA), told TASS.

"This platform serves as a universal integrator for virtually the entire data stream generated by the combat use of various types of unmanned systems, which enables comprehensive training of neural networks, the establishment of unified standards and approaches for the operation of such systems, and their centralized development and deep learning. In doing so, we are laying the groundwork for seamless, end-to-end platforms for deploying various types of strike-reconnaissance drones, integrating them into a unified digital ecosystem, and ensuring maximum efficiency and coordination in their use. In this sense – beyond reconnaissance analytics and target designation – we are addressing the task of enhancing the autonomy of our army of the future, with an eye toward the deployment of autonomous robotic systems," the expert said.

According to him, the neural network training system could eventually be used to develop sixth-generation aircraft. "The platform’s scope will extend beyond the atmosphere to include aerospace, facilitating the training of next-generation systems designed to operate in low-Earth orbit and perform tasks such as anti-satellite operations. We are talking about sixth-generation systems – including hypersonic ones – capable of securing our airspace and, if necessary, neutralizing adversary spacecraft carrying various types of weaponry," Stepanov added.

On October 12, the Russian Defense Ministry announced that its specialists created a service that provides developers of unmanned systems with access to live recording data of drones’ combat deployment for the training of military-purpose neural networks. The new service provides developers of unmanned systems with access to a specialized repository containing prepared video footage of the results of the use of FPV drones and other copter drones and fixed-wing unmanned aerial vehicles.

In addition, the ministry’s specialists have created a unified data recognition classifier for machine learning AI systems, which allows developers to work with the database efficiently, in a structured manner, and according to a unified plan. The results of the work conducted by developers of military-purpose unmanned systems are uploaded back to the service, where the information is checked for completeness and compliance with the classifier format, enabling the Defense Ministry to use this dataset to create and train its neural networks.