Primjena strojnog učenja za klasifikaciju prijetnji iz zapisa o kibernetičkoj sigurnosti

undergraduate thesis

• Title (english): Application of machine learning to threat classification from cybersecurity records
• Author: Mirta Medak
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Marko Barišić (član povjerenstva)
• Committee member: Domagoj Pluščec (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-05, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Kibernetički napadi predstavljaju veliki problem za sve korisnike računalnih sustava. Procjena
rizika kibernetičkih prijetnji može biti dugačak i skup proces. Za procjenu opasnosti
od prijetnje koristi se ocjena Common Vulnerability Scoring System (CVSS). CVSS ranjivostima
dodjeljuju stručnjaci. Uz pomoć metoda obrade prirodnog jezika moguće je predvidjeti
CVSS iz opisa ranjivosti. Regresija potpornih vektora i BERT regresor korišteni su
za izravno predviđanje osnovne komponente CVSS-a iz opisa. Regresijski modeli pokazali
su se dovoljno uspješnima s obzirom na postojeće rezultate, ali postoji puno prostora za
napredak. S druge strane, moguće je predvidjeti parametre CVSS-a u rasponu točnosti od
0,90-0,95 koristeći BERT klasifikacijski model. Ovaj rezultat pokazao je barem jedan korak
prema automatiziranom predviđanju CVSS-a.

Abstract (english)

Cyber attacks are a problem for all users of computer systems. Prioritization and evaluation
of threats can be difficult and costly. The Common Vulnerability Scoring System (CVSS)
score is used to assess the danger of a threat. CVSS is assigned by experts. With help of NLP
methods, it is possible to predict CVSS score from the description of a vulnerability. Support
Vector Regression and BERT Regressor were used to directly predict the base component
of CVSS. The models performed well compared to already existent methods, but there is
space for improvement. On the other hand, it is possible to predict the score metrics in the
accuracy range of 0.90-0.95 using the BERT classification model. This would be at least one
step towards automated prediction of CVSS.

• Keywords: obrada prirodnog jezika; strojno učenje; CVE zapisi; CVSS; regresija potpornih vektora; BERT; transformeri; BERT za regresiju; BERT za klasifikaciju
• Keywords (english): natural language processing; machine learning; Common Vulnerability and Exposure records; CVSS score; Support Vector Regression; BERT; transformers; BERT Regressor; BERT Classification
• Language: english
• URN:NBN: urn:nbn:hr:168:490781
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-15 10:49:11