Incorporating Temporal Discretization into Naïve Bayes Learning for Classifying a Driver’s Cognitive Load

摘要

This study incorporates a novel temporal discretizationnmethod into Naïve Bayes Learning for classifying a driver’sncognitive load. We regard the temporal value, tendency, andnstability as important features for classifying a driver’s state.nTime series segmentation and K-means clustering are appliednto discretize time-series sensor data. We then apply the NaïvenBayes Classifier model to classify the driver’s state. Wenpresent the classification problem of the car driver’s state. Wenneed a function in the in-vehicle information service thatnjudges the user’s cognitive load and define the driver’sncognitive load based on the driving situation. The experimentnindicated that the best accuracy was 86.0 percent, with annoverall accuracy of 73.5 percent. Our proposed methodnexcelled over conventional methods; however, furthernaccuracy improvement is necessary.