ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection
Driver distraction detection (3D) is essential in improving the efficiency and safety of transportation systems. Considering the requirements for user privacy and the phenomenon of data growth in real-world scenarios, existing methods are insufficient to address four emerging challenges, i.e., data...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-04-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/11/8/1867 |
| _version_ | 1797604498747686912 |
|---|---|
| author | Zihan Guo Linlin You Sheng Liu Junshu He Bingran Zuo |
| author_facet | Zihan Guo Linlin You Sheng Liu Junshu He Bingran Zuo |
| author_sort | Zihan Guo |
| collection | DOAJ |
| description | Driver distraction detection (3D) is essential in improving the efficiency and safety of transportation systems. Considering the requirements for user privacy and the phenomenon of data growth in real-world scenarios, existing methods are insufficient to address four emerging challenges, i.e., data accumulation, communication optimization, data heterogeneity, and device heterogeneity. This paper presents an incremental and cost-efficient mechanism based on federated meta-learning, called ICMFed, to support the tasks of 3D by addressing the four challenges. In particular, it designs a temporal factor associated with local training batches to stabilize the local model training, introduces gradient filters of each model layer to optimize the client–server interaction, implements a normalized weight vector to enhance the global model aggregation process, and supports rapid personalization for each user by adapting the learned global meta-model. According to the evaluation made based on the standard dataset, ICMFed can outperform three baselines in training two common models (i.e., DenseNet and EfficientNet) with average accuracy improved by about 141.42%, training time saved by about 54.80%, communication cost reduced by about 54.94%, and service quality improved by about 96.86%. |
| first_indexed | 2024-03-11T04:47:09Z |
| format | Article |
| id | doaj.art-e916cb90b6b94394bfaa42ada966e63b |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-11T04:47:09Z |
| publishDate | 2023-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-e916cb90b6b94394bfaa42ada966e63b2023-11-17T20:17:42ZengMDPI AGMathematics2227-73902023-04-01118186710.3390/math11081867ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction DetectionZihan Guo0Linlin You1Sheng Liu2Junshu He3Bingran Zuo4School of Intelligent Systems Engineering, Sun Yat-Sen University, Shenzhen 518107, ChinaSchool of Intelligent Systems Engineering, Sun Yat-Sen University, Shenzhen 518107, ChinaSchool of Intelligent Systems Engineering, Sun Yat-Sen University, Shenzhen 518107, ChinaSchool of Intelligent Systems Engineering, Sun Yat-Sen University, Shenzhen 518107, ChinaRehabilitation Research Institute of Singapore, Nanyang Technological University, Singapore 639798, SingaporeDriver distraction detection (3D) is essential in improving the efficiency and safety of transportation systems. Considering the requirements for user privacy and the phenomenon of data growth in real-world scenarios, existing methods are insufficient to address four emerging challenges, i.e., data accumulation, communication optimization, data heterogeneity, and device heterogeneity. This paper presents an incremental and cost-efficient mechanism based on federated meta-learning, called ICMFed, to support the tasks of 3D by addressing the four challenges. In particular, it designs a temporal factor associated with local training batches to stabilize the local model training, introduces gradient filters of each model layer to optimize the client–server interaction, implements a normalized weight vector to enhance the global model aggregation process, and supports rapid personalization for each user by adapting the learned global meta-model. According to the evaluation made based on the standard dataset, ICMFed can outperform three baselines in training two common models (i.e., DenseNet and EfficientNet) with average accuracy improved by about 141.42%, training time saved by about 54.80%, communication cost reduced by about 54.94%, and service quality improved by about 96.86%.https://www.mdpi.com/2227-7390/11/8/1867federated learningmeta-learningincremental federated meta-learningdriver distraction detectionICMFed |
| spellingShingle | Zihan Guo Linlin You Sheng Liu Junshu He Bingran Zuo ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection Mathematics federated learning meta-learning incremental federated meta-learning driver distraction detection ICMFed |
| title | ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection |
| title_full | ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection |
| title_fullStr | ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection |
| title_full_unstemmed | ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection |
| title_short | ICMFed: An Incremental and Cost-Efficient Mechanism of Federated Meta-Learning for Driver Distraction Detection |
| title_sort | icmfed an incremental and cost efficient mechanism of federated meta learning for driver distraction detection |
| topic | federated learning meta-learning incremental federated meta-learning driver distraction detection ICMFed |
| url | https://www.mdpi.com/2227-7390/11/8/1867 |
| work_keys_str_mv | AT zihanguo icmfedanincrementalandcostefficientmechanismoffederatedmetalearningfordriverdistractiondetection AT linlinyou icmfedanincrementalandcostefficientmechanismoffederatedmetalearningfordriverdistractiondetection AT shengliu icmfedanincrementalandcostefficientmechanismoffederatedmetalearningfordriverdistractiondetection AT junshuhe icmfedanincrementalandcostefficientmechanismoffederatedmetalearningfordriverdistractiondetection AT bingranzuo icmfedanincrementalandcostefficientmechanismoffederatedmetalearningfordriverdistractiondetection |