Institutional Repository of Vilnius University
Įterptinių instrukcijų atakų (XSS) aptikimas naudojant dirbtinius neuroninius tinklus /
Abstract
dc:descriptionCross-site-scripting (XSS) attacks are one of the most common cyberattacks against web applications. According to often in literature cited cybersecurity organization "OWASP", injection attacks (XSS included) take third place in vulnerability occurrence. XSS is usually classified into the three groups – reflected, stored and DOM. It is possible to find vulnerabilities without using additional software – using XSS polyglot, unit testing. However, detecting cross-site-scripting vulnerabilities using these methods could be difficult and ineffective in larger websites. Therefore, software that could ease this job was created. In order to prevent a cyberattack, preventive methods are being used such as encoding, HTML sanitization, frameworks, web application firewall. However, these methods are not always giving satisfactory results. Therefore, alterantive methods such as usage of artificial neural networks, were considered. In this bachelor’s work, a convolutional neural network model, that is able to detect XSS attacks, was created. The conducted experiments determined the optimal epoch number, batch size, learning rate, weight decay rate, dropout rate and best vectorization, tokenization methods. Lastly, a created model was implemented in logging. A website was created that sends requests to "Seq" server, "Seq" server logs the requests, groups possible attacks and sends e-mails about them every 10 minutes.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Baliukynas, Domantas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192063354&prefLang=en_US