FINDING · DETECTION

A novel LSTM + CNN architecture achieves 95% service classification accuracy on a real-world ISP and mobile operator dataset using less raw traffic and fewer parameters than prior approaches, outperforming the state-of-the-art by nearly 50% fewer false classifications.

From 2022-akbari-traffic-classification-increasingly — Traffic classification in an increasingly encrypted web · Abstract · 2022 · Communications of the ACM

Implications

Even with reduced traffic volume, ML classifiers can achieve near-SOTA accuracy — circumvention tools cannot rely on limiting the number of observable packets as a defense; flow-level shaping must address statistical distributions, not just packet count.
The 50% false-classification reduction means ISP-deployed classifiers are becoming precise enough to avoid false positives that would block legitimate traffic, lowering the political cost of deploying them.