<ul><li>MIETT (Multi-Instance Encrypted Traffic Transformer) is a model for classifying encrypted network traffic.</li><li>It adopts a multi-instance approach, treating each packet as a distinct instance within a larger bag representing the entire flow.</li><li>MIETT utilizes Two-Level Attention (TLA) layers to capture both token-level and packet-level relationships effectively.</li><li>The model incorporates novel pre-training tasks, PRPP and FCL, to enhance its understanding of temporal and flow-specific dynamics.</li></ul>

MIETT: Multi-Instance Encrypted Traffic Transformer for Encrypted Traffic Classification

Discover more