Usability of the Artificial Intelligence and Modern Techniques for Securing Computer Systems

Bc. David MALANÍK

Disertační práce

Abstract:

This thesis is aimed to practical usability of the neural network in computer security application. The main point of this thesis is focused on the small part of neural network inside the user verification process. The neural network represented some intelligent system which might be adapted to specific user characteristic and might be usable for the smart reconstruction and identification users. Other part of this work deals with the next potentially usability of smart neural network. One of these parts includes the cipher algorithms based on the neural network. Other one represented the new angle of view onto user verification. The commonly base method use only one-time user verification. But the systems included some intelligent behaviour, might periodically checking the user identity. Obviously not by providing the username and some password, but by processing the statistics analyses of current user. These systems are able to detect the user- switching without the correct logout and login procedure.

Abstract:

Tato pra?ce je zame?r?ena na moz?nosti vyuz?iti? neuronovy?ch si?ti? v oblasti zabezpec?eni? poc?i?tac?ovy?ch syste?mu?. Hlavni? c?a?st te?to pra?ce je zame?r?ena na vyuz?itelnost neuronovy?ch si?ti? pro identifikaci uz?ivatelu?. Neuronove? si?te? pr?edstavuji? inteligentni? syste?m, ktery? je adaptovatelny? na specificke? vlastnosti uz?ivatelu? a lze jej pouz?i?t i pro inteligentni? rekonstrukci a identifikaci uz?ivatelu?. V dals?i? c?a?sti pra?ce je popsa?no jine? moz?ne? vyuz?iti? neuronove? si?te?, jedna? se o moz?nost vyuz?iti? neuronove? si?te? jako cryptograficke?ho apara?tu pro zabezpec?eni komunikace. Dals?i? c?a?st ukazuje novy? pohled na verifikaci uz?ivatelu?. Be?z?ne? metody pro ove?r?ova?ni? uz?ivatelu? jsou zaloz?eny na jednora?zove? identifikaci uz?ivatele. Navrz?eny? syste?m ale obsahuje inteligentni? rutinu, ktera? periodicky ove?r?uje identitu uz?ivatele. Toto over?eni? ale neni? realizova?no pomoci? dotazu na jme?no a heslo ale vyuz?i?va? statisticke?ho vzorce chova?ni? uz?ivatele. Syste?m je tedy schopen detekovat zme?nu uz?ivatele bez korektni?ho odhla?s?eni? sta?vaji?ci?ho uz?ivatele a pr?ihla?s?eni? nove?ho.

Keywords

bezpec?nost poc?i?tac?ovy?ch syste?mu?, ume?le? neuronove? si?te?, identity manager, autentizace uz?ivatelu?, biometricka? identifikace, kryptografie

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 28. 7. 2011

Zveřejnit od: 28. 7. 2011

Obhajoba závěrečné práce

• Obhajoba proběhla 28. 11. 2011

Univerzita Tomáše Bati ve Zlíně

Fakulta aplikované informatiky

Doktorský studijní program / obor:
Inženýrská informatika / Inženýrská informatika