RESEARCH ON ATTACK DETECTION MODELS FOR DISTRIBUTED SYSTEMS USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

With the exponential growth of information that is available to mankind, the corresponding need to process the growing volumes of data also grows. Distributed systems are great for solving such problems, but in turn have certain disadvantages, among them - susceptibility to interference attacks. Such attacks can be prevented by detecting them in time and taking appropriate actions necessary for protection. This article examines the approaches and methods by which a good level of attack detection can be achieved, to test the hypotheses, experiments were performed on self-developed models of convolutional neural networks.

The available articles on this topic were reviewed and a conclusion was drawn about the rapid development of this direction and the need to continue it.

By combining experiments with different models of neural networks, a variety of processing of input data was performed in order to increase the accuracy and quality of intervention detection.The results of the conducted experiments and the corresponding preparations for them are analyzed in detail. As a result, this article provides important information about effective methods and approaches to improve attack detection accuracy and distributed system security, respectively.

Further directions for the development of this topic are also given, which are interesting and important for new research