< Back to previous page

Project

Protection of security solutions that rely on machine learning

A growing number of security solutions and building blocks rely on the promising value of machine learning. Major cases in this research include authentication, intrusion detection, malware analysis and the processing of audit trails. The application of machine learning techniques in this specific context must be thoroughly investigated as security solution inherently face challenges imposed by adversaries. This thesis will survey and investigate threats and countermeasures for strong and weak attack models, and research, develop and evaluate methods that strengthen machine learning methods and environments to become secure and robust in such an adversarial setting. Prototypical solutions will be applicable in a middleware context.

Date:6 May 2019 →  6 May 2023
Keywords:Computer Security, Adversarial Machine Learning
Disciplines:Computer system security
Project type:PhD project