A Character Prediction Approach in a Security Context using a Recurrent Neural Network
| Autor: | Valentin-Alexandru Vladuta, Andrei-Marius Avram, Ana-Maria Ghimes |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Web server
Point (typography) Computer science business.industry Character (computing) 05 social sciences Big data 050801 communication & media studies 02 engineering and technology Security context computer.software_genre Computer security 0508 media and communications Recurrent neural network 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Language model business computer |
| Zdroj: | 2018 International Symposium on Electronics and Telecommunications (ISETC). |
| DOI: | 10.1109/isetc.2018.8584007 |
| Popis: | Nowadays, cybersecurity focuses on detecting intrusions and anomalies with the aid of automated methods, exploratory visual analysis systems or real-time monitoring with dynamic visual representations. Current research in cybersecurity is not yet at the point where analysts can rely on systems for assuring security. Being in a Data Era, the amount of gathered data can play opposite roles: on one side, to serve the security of an organization, but on the other, to pose security threats. Our paper aims to present a technical implementation of a recurrent neural network that predicts characters in a security context. The end goal of the proposition is to create an application for adding security to a web server by predicting log entries before an actual attack. If such a server would get a certain entry pattern, it will predict a new log before an actual attack. Results demonstrate that character prediction can be achieved with 40% accuracy. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|