Mariconti, Enrico; (2019) You Shall Not Pass! Measuring, Predicting, and Detecting Malware Behavior. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Expmain.pdf - Accepted Version Download (3MB) | Preview

Abstract

Researchers have been fighting malicious behavior on the Internet for several decades. The arms race is far from being close to an end, but this PhD work is intended to be another step towards the goal of making the Internet a safer place. My PhD has focused on measuring, predicting, and detecting malicious behavior on the Internet; we focused our efforts towards three different paths: establishing causality relations into malicious actions, predicting the actions taken by an attacker, and detecting malicious software. This work tried to understand the causes of malicious behavior in different scenarios (sandboxing, web browsing), by applying a novel statistical framework and statistical tests to determine what triggers malware. We also used deep learning algorithms to predict what actions an attacker would perform, with the goal of anticipating and countering the attacker’s moves. Moreover, we worked on malware detection for Android, by modeling sequences of API with Markov Chains and applying machine learning algorithms to classify benign and malicious apps. The methodology, design, and results of our research are relevant state of the art in the field; we will go through the different contributions that we worked on during my PhD to explain the design choices, the statistical methods and the takeaways characterizing them. We will show how these systems have an impact on current tools development and future research trends.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item