Abstract: This research proposes a self-supervised anomaly detection framework tailored for encrypted network flows, leveraging Gated Recurrent Unit (GRU) models trained with sequence reconstruction ...