Welcome to this workshop
Machine Learning (ML) models have been shown to be vulnerable to adversarial examples designed to fool ML models to classify them as benign rather than malicious. This has led to several research efforts geared towards the exploration of adversarial learning in a bid to stay ahead of attackers. A problem with this approach though it that the adversarial samples generated are not often tested to ensure that they remain executable and retain their malicious functionality. Thus, the need for more studies/discussion groups/workshops in this area and hence the workshop will be on the executability, and malicious retention of adversarial malware samples generated using adversarial learning. The purpose of the workshop is to bring together some researchers within Scottish Universities and beyond who currently work on adversarial malware generation to discuss how to preserve the executability and malicious nature of samples generated through adversarial learning. The motivation being that the focus of the community is often on generating samples and not necessarily on whether they remain executable and malicious which are quite key as there is no point creating malware mutants that are non-executable and non-malicious to serve as training data to improve their classification.
