Training neural networks for construction of informatics offender profile

The increase of electronic data processing and transmission via computer networks has laid more and more emphasis on the security of these networks. The people attempting to breach network security may belong to a wide range of categories, committing more or less serious offences. Conventional means do not offer adequate protection for internet transmitted threats, and there is no method or technology that can ensure by itself an improvement of this situation. By contrast, a multilayer, complete set of techniques might - if not eliminate - at least diminish the negative effect of these threats. Therefore, a new approach appears to be necessary to ensure the informatics security. By identification of the profile of the cybercriminal, the psychology of cyber criminology directs its attention towards the application of the physical, psychological, social and mental characteristics, as well as towards the evidence of the cybercrimes. The present paper presents a system used to identify the profile of the cybercriminal through the involvement of the feedforward neural networks using genetic algorithms. This system is based upon the environmental offences and variables that reflect the evidence of the cybercrime, as well as its circumstances, and its outcome is the description of the characteristics of the psychological and behavioural profile.