Jet Infosystems constructed the neuronet revealing network attacks for Fund of perspective researches

2019: Solution development on preventive identification of network attacks

The Jet Infosystems company announced on June 18, 2019 development of the neuronet allowing to reveal network attacks within the competition organized by the Fund of Perspective Researches (FPR).

Search of the innovation ideas and technological solutions on application AI/ML in problems of providing was the purpose of tender information security.

By rules of tender of Fund of perspective researches it was required to select optimal approach from detection of the disguised network attacks. It was for this purpose offered to develop the software solution for the analysis of two sets of network traffic: for training and for solution quality evaluation. The itemized statement on the detected threats at a speed approximate to real time had to become result of a program runtime.

In work on a competitive task the command Jet Infosystems decided to apply methods of machine learning as most productive means to detection of difficult patterns and network anomalies. As a result the developed AI solution with a high accuracy (about 90%) identified a class of the attack, number of network packets and the operated vulnerability of software if like that was present.

We carried out the behavioural analysis of five classes of the attacks, and then built neural network for detection of vidoizmenny network invasions — told Anna Oleynikova, the head of solutions on detection of network invasions Jet Infosystems. — Machine learning for us — strategic direction of development, and we understand that modern security concerns cannot be solved without the aid of artificial intelligence.

Further the solution developed by Jet Infosystems will be used in the FPI project in information security field. In its framework it will be required to create technology and a demonstration sample which, being on perimeter or in network, will be able to identify the normal and disguised network attacks quickly to warn and detect harmful actions of malefactors at early stages of their preparation.