Transformer Encoders Achieve 0.93 F1 in ADS Classification
August 3, 2026
Transformer encoders outperform GRU (0.92 F1) and LSTM (0.90 F1) models when classifying automated driving systems like Tesla Autopilot and Cadillac Super Cruise using vehicle telematics. The models provide a method for independently verifying active driver assistance levels for safety and insurance purposes.
HOW THIS AFFECTS YOU
●
builderYou can implement these encoders to build third-party monitoring systems for automated driving.
●
researcherYou can use these macro F1-score baselines for sequence-based vehicle telematics classification.