Application of Neural Networks for Data Protection Research

Abstract

About 80% of computer viruses are distributed by e-mail. Approximately 10-15 billion US dollars are spent annually for theliquidation of their outcomes. With every year, the quantity of computer viruses is increasing, and the contemporary systems are not ableto stop this newly appearing damaging software. In this work, the attempts are taken to design a system that operates basing on theprinciples of biological epidemiology arrest and is controlled by the neural network. Conclusions are made in this work that it isadvisable to install antiviral systems in common e-mail servers instead of the computers of final users, and also to group the networkinto the smallest possible segments. System design has been carried out and experiment of classification of infected and not infectedmessages using neural networks has been done. The experiment has shown the flexibility and prospects of this technology. Ill. 6, bibl. 9(in English, summaries in Lithuanian, English, Russian).