The purpose of the research is to research and implement new functionality of the Computational segment of the Prototype, to solve problems at the macro level. The found characteristics of the UDS loading can be used to calculate a number of transport indicators, as well as to forecast the volume of emissions of pollutants and noise from road transport and other purposes.

Within the framework of the project, a mathematical model of the customer's logistics system, an optimization model for the task of minimizing the cost of cargo transportation, prototypes of algorithms for the numerical solution of the optimization problem of minimizing the cost of cargo transportation, taking into account restrictions on throughput and available time windows, will be developed.

The main goal of the project is to develop new practical modifications of vertical federated learning algorithms that the Employer already has. New approaches to the task of vertical federated learning will be developed, different from those that are present in the literature. The final modifications and algorithms will be extended from classical machine learning models (which the Employer is currently using) to more complex ones, for instance, neural networks.

Bandwidth, speed, latency, and resistance to interference are all important characteristics of modern Internet networks. A team of young scientists from MIPT will deal with quality metrics for networks, as well as algorithms for their optimization. The developments will allow us to figure out how to lay a fiber-optic line in the most successful way, where it is more optimal to put a switch and so forth.

The technique developed based on the results of the project will allow you to quickly train the best and most accurate models of deep/in-depth learning without expensive computing devices. This approach will give access to advanced results of deep/in-depth learning not only to large companies with sufficient computing resources, but also to a significant number of users.

The teamwork of young scientists from MIPT demonstrated not only the possibility of creating a package in Russia in the coming years that would be competitive with Western models, but would also surpass them in such important indicators as reliability (robustness) and speed. The latter is very important, because the main purpose of such packages for transport designers is to decide where to build new roads, make interchanges, dedicated lanes, etc.