Details
| Original language | English |
|---|---|
| Title of host publication | 2020 IEEE 23rd International Conference on Intelligent Transportation Systems, ITSC 2020 |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| ISBN (electronic) | 9781728141497 |
| ISBN (print) | 978-1-7281-4150-3 |
| Publication status | Published - 2020 |
| Event | 23rd IEEE International Conference on Intelligent Transportation Systems, ITSC 2020 - Rhodes, Greece Duration: 20 Sept 2020 → 23 Sept 2020 |
Abstract
Recent studies on machine learning based driver identification have shown that leveraging deep neural networks to learn latent features from vehicle sensor data boosts the performance of the models to high levels of accuracies. However, models produced by deep neural networks are difficult to explain and the interpretability of their results is limited. Consequently, the reliability of the learned models heavily depends on the amount, quality and diversity of the training dataset as well as on the validation scenarios. In this work, we evaluate state-of-the-art deep learning networks for driver identification using a very rich dataset of more than 395, 000 kilometres of different driving scenarios and environmental conditions (e.g. route, vehicle, traffic, weather) collected over two years. It turns out that the neural networks achieve high accuracy levels when training and testing on the same type of driving conditions. However, accuracy drops and importance of individual signals varies when testing on different driving conditions, although all best practices like stratification and cross validation have been applied. Our findings suggest that relying solely on the vehicle sensor data without taking the contextual information about driving conditions into account is not a practical solution.
ASJC Scopus subject areas
- Computer Science(all)
- Artificial Intelligence
- Decision Sciences(all)
- Decision Sciences (miscellaneous)
- Decision Sciences(all)
- Information Systems and Management
- Mathematics(all)
- Modelling and Simulation
- Social Sciences(all)
- Education
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
2020 IEEE 23rd International Conference on Intelligent Transportation Systems, ITSC 2020. Institute of Electrical and Electronics Engineers Inc., 2020. 9294439.
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - On the Importance of Contextual Information for Building Reliable Automated Driver Identification Systems
AU - Zeng, Li
AU - Al-Rifai, Mohammad
AU - Chelaru, Sergiu
AU - Nolting, Michael
AU - Nejdl, Wolfgang
N1 - Funding Information: ACKNOWLEDGMENT The authors affiliated with Volkswagen AG have received funding from Volkswagen AG to conduct this research.
PY - 2020
Y1 - 2020
N2 - Recent studies on machine learning based driver identification have shown that leveraging deep neural networks to learn latent features from vehicle sensor data boosts the performance of the models to high levels of accuracies. However, models produced by deep neural networks are difficult to explain and the interpretability of their results is limited. Consequently, the reliability of the learned models heavily depends on the amount, quality and diversity of the training dataset as well as on the validation scenarios. In this work, we evaluate state-of-the-art deep learning networks for driver identification using a very rich dataset of more than 395, 000 kilometres of different driving scenarios and environmental conditions (e.g. route, vehicle, traffic, weather) collected over two years. It turns out that the neural networks achieve high accuracy levels when training and testing on the same type of driving conditions. However, accuracy drops and importance of individual signals varies when testing on different driving conditions, although all best practices like stratification and cross validation have been applied. Our findings suggest that relying solely on the vehicle sensor data without taking the contextual information about driving conditions into account is not a practical solution.
AB - Recent studies on machine learning based driver identification have shown that leveraging deep neural networks to learn latent features from vehicle sensor data boosts the performance of the models to high levels of accuracies. However, models produced by deep neural networks are difficult to explain and the interpretability of their results is limited. Consequently, the reliability of the learned models heavily depends on the amount, quality and diversity of the training dataset as well as on the validation scenarios. In this work, we evaluate state-of-the-art deep learning networks for driver identification using a very rich dataset of more than 395, 000 kilometres of different driving scenarios and environmental conditions (e.g. route, vehicle, traffic, weather) collected over two years. It turns out that the neural networks achieve high accuracy levels when training and testing on the same type of driving conditions. However, accuracy drops and importance of individual signals varies when testing on different driving conditions, although all best practices like stratification and cross validation have been applied. Our findings suggest that relying solely on the vehicle sensor data without taking the contextual information about driving conditions into account is not a practical solution.
UR - http://www.scopus.com/inward/record.url?scp=85099639548&partnerID=8YFLogxK
U2 - 10.1109/ITSC45102.2020.9294439
DO - 10.1109/ITSC45102.2020.9294439
M3 - Conference contribution
AN - SCOPUS:85099639548
SN - 978-1-7281-4150-3
BT - 2020 IEEE 23rd International Conference on Intelligent Transportation Systems, ITSC 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 23rd IEEE International Conference on Intelligent Transportation Systems, ITSC 2020
Y2 - 20 September 2020 through 23 September 2020
ER -