Mgr. Olha Jurečková

Mgr. Olha Jurečková

Mgr. Tomáš Rabas

Department of Information Security

This thesis presents an analysis of the randomness of the Dumbo instance of the Elephant cipher. The author used various statistical tests, including tests from NIST's Statistical Test Suite, the d-monomial test and cube testers. The results showed that Dumbo passed the majority of the tests from NIST's STS with only a few failures that were not considered significant. The d-monomial test also showed no bias in any particular keystream bit. The most significant finding of this work was the discovery that Dumbo had difficulty passing the balance test from the family of cube testers as some keystream bits only passed every other test case and many cubes resulted in a pass rate of around 20 %. The thesis concludes that Dumbo demonstrates good randomness properties overall but highlights the need for further investigation of this cipher to identify any potential major issues.

Thesis on DSpace

Mgr. Olha Jurečková

Ing. Matouš Kozák

Department of Information Security

This paper examines malware classification using self-organizing maps. Detection using artificial intelligence is one way to effectively combat the spread of malware, and the ability of self-organizing maps to organize data into clusters can aid in classification. This paper summarizes the theoretical basis and the development of self-organization maps so far. Afterward, a test map is trained on a publicly available dataset with parameters selected for this type of data. In the last section, the measured results are compared with other machine learning methods.

Thesis on DSpace

Mgr. Olha Jurečková

Ing. Filip Kodýtek, Ph.D.

Department of Information Security

Recently machine learning (ML) techniques become very popular in many areas, including natural language processing, voice/image recognition, etc. The goal of this master thesis is to create a technique of malware detection by using image visualization. As a method of gaining information from converted malware samples classification using Convolutional Neural Network (CNN) detection was chosen. The malware sample represented as byte array is first converted to gray-scale or RGB image and then fed as input to CNN.

Thesis on DSpace