Zivakin V. Information technology for recognition of aerial survey data using multidimensional representations

The dissertation work is devoted to the study of neural network processing of aerial survey training set data, recognition classification processes and set data analysis, manipulation of these data using multidimensional representations and their impact on the results of solving classification problems. Today, the active phase of the war continues in Ukraine, which is also called the "war of drones". Unmanned aerial vehicles have proven to be an effective, cheap, safe and universal technical tool for solving tasks of various types and levels (from tactical to strategic). Ensuring a quantitative and qualitative advantage in such means is one of the factors for achieving victory and strengthening the country's further defense capability. The main tasks that are solved when using UAVs: navigation along the optical channel (in the conditions of constant action of EW devices), identification and support of targets. In the context of solving these problems, the component of recognizing terrain objects and target objects is essential, for which neural network processing is used. The results of the work correspond to the request of the following orders of the Cabinet of Ministers: o No. 600 "Some issues of the development of critical technologies in the field of production of weapons and military equipment" o No. 1556-r "On the approval of the Concept of the development of artificial intelligence in Ukraine". That resolution: o No. 256 - "On the implementation of an experimental project on the implementation of defense procurement of domestically produced unmanned systems and means of radio-electronic warfare." o No. 476 - "On approval of the list of priority thematic areas of scientific research and scientific and technical development for the period until December 31 of the year following the termination or abolition of martial law in Ukraine", the list of which includes the following items: The purpose of the work is the development of information technology for digital image recognition based on the developed model for evaluating the distribution of data representations of the aerial survey training set in multidimensional space and using the proposed pre-processing methods. Therefore, the dissertation research has significant scientific and practical value, because its results will allow to improve the results of solving classification problems and to introduce a new model of distribution estimation for training sets.