AI-based user authentication reinforcement by continuous extraction of behavioral interaction features

Abstract In this work, we conduct an experiment to analyze the feasibility of a continuous authentication method based on monitorization users’ activity verify their identities through specific user profiles modeled via Artificial Intelligence techniques. order experiment, custom application was developed gather records in guided scenario where some predefined actions must be completed. This dataset has been anonymized and will available community. Additionally, public also used for benchmarking purposes so that our techniques could validated non-guided scenario. Such data were processed extract number key features train three different techniques: Support Vector Machines, Multi-Layer Perceptrons, Deep Learning approach. These demonstrated perform well both scenarios, being able authenticate users effective manner. Finally, rejection test conducted, system proposed tested using weighted sliding windows, impostor detected real environment when legitimate session is hijacked.