Modified TSception for Driver Drowsiness and Mental Workload Detection
Analysis
This paper introduces a modified TSception architecture for EEG-based driver drowsiness and mental workload assessment. The key contributions are a hierarchical architecture with temporal refinement, Adaptive Average Pooling for handling varying EEG input dimensions, and a two-stage fusion mechanism. The model demonstrates comparable accuracy to the original TSception on the SEED-VIG dataset but with improved stability (reduced confidence interval). Furthermore, it achieves state-of-the-art results on the STEW mental workload dataset, highlighting its generalizability.
Key Takeaways
- •Proposes a modified TSception architecture for EEG-based driver drowsiness and mental workload detection.
- •Introduces a hierarchical architecture with temporal refinement and Adaptive Average Pooling.
- •Achieves comparable accuracy to the original TSception with improved stability on the SEED-VIG dataset.
- •Demonstrates state-of-the-art results on the STEW mental workload dataset, highlighting generalizability.
“The Modified TSception achieves a comparable accuracy of 83.46% (vs. 83.15% for the original) on the SEED-VIG dataset, but with a substantially reduced confidence interval (0.24 vs. 0.36), signifying a marked improvement in performance stability.”