USE OF NEURAL NETWORKS FOR DETECTION AND RECOGNITION OF THE ANOMALIES IN ENTERPRISE CORPORATIVE INFORMATION SYSTEM

Abstract

The neural networks structure using for task solving of information defense are discussed. The choice of attribute and metadata of executing files with two states (clean and with viruses) which used for multilevel perseptron teaching are built. The teaching was realized within SPSS Statistics - program of IBM Company. After the teaching of neural network the efficiently its working with the control choice of executing files was determined. The relative value of files classification was 5 %, that it is the good result. The file choice must be greater and viruses more variables within the use such approach for large enterprise corporative information systems.

Keywords

• neural network structure
• information defense
• viruses file
• teaching of neuron net
• corporative information system