Adversarial Attacks on Time Series Models: A Systematic Literature Review

Time series classification (TSC) is a transversal task that nowadays finds many applications in many different fields, including healthcare and safety in industrial environments. It is clear that for many of these applications, a misclassified time series can have dangerous or harmful consequences and, therefore, it is fundamental to understand the weaknesses of the TSC models in order to define models as robust as possible against adversarial attacks. The aim of this project is to conduct a Systematic Literature Review (SLR) of the studied and empirically validated adversarial attacks on TSC models.