|Table of Contents|

Federated heterogeneous model and algorithm for personal travel recommendation(PDF)

《交通运输工程学报》[ISSN:1671-1637/CN:61-1369/U]

Issue:
2023年05期
Page:
253-263
Research Field:
交通信息工程及控制
Publishing date:
2023-11-10

Info

Title:
Federated heterogeneous model and algorithm for personal travel recommendation
Author(s):
YOU Lin-lin1;  HE Jun-shu1;  CHEN Kun-xu1;  HE Jia-qi1;  YUAN Shao-xin2;  ZHAO Juan-juan3;  CAI Ming1
(1. School of Intelligent Systems Engineering, Sun Yat-Sen University, Shenzhen 518107, Guangdong, China; 2. School of Information Engineering, Chang'an University, Xi'an 710064, Shaanxi, China; 3. College of Resource Environment and Tourism, Capital Normal University, Beijing 100048, China)
Keywords:
intelligent transportation;  travel recommendation;  federated learning;  mixed logit model;  privacy protection;  personalization
PACS:
U491.1
DOI:
10.19818/j.cnki.1671-1637.2023.05.018
Abstract:
To achieve personal travel recommendations by considering both the preference heterogeneity and data privacy, based on the model parametric aggregation and distributed training supported by the federated learning(FL)computing paradigm, a federated mixed Logit(FMXL)model was proposed by decoupling the standard mixed Logit model to separate the parameter estimation of local individual preferences and global population differences. In order to eliminate the dependence of the model on the original data, two federated Gibbs sampling algorithms, to be standardized or aggregated, were designed, to achieve the hierarchical estimation of the model through the interaction of local and global parameters. The proposed model and algorithm were validated for offline and online travel recommendation scenarios based on Swiss Metro data. Analysis results show that for offline scenario, the FMXL model based on two federated Gibbs sampling algorithms increase the log-likelihood value by 157.8 and 153.2, and the prediction rate by 12.3% and 12.1%, respectively, compared with the standard multinomial Logit model. In addition, the computation time reduce by 64.2% and 76.9%, respectively, and the communication times both reduce by 86.2% compared with the mixedLogit model based on the centralized Gibbs sampling algorithm. For the online scenario, both the log-likelihood value and the prediction rate of the FMXL model show an increasing trend, and the computation and communication times of the whole estimation process are lower than those of the standard mixed Logit model. Overall, with the data privacy as the precondition, the federated training of the MXL model ensures the accuracy of travel recommendations and effectively enhances the timeliness of travel recommendations by fully utilizing the idle computing power on the user side, reflecting the high adaptability and scalability of the proposed model and algorithms. In addition, the intelligent progress of the transportation system can be promoted effectively by personal travel recommendations based on the FMXL model. 3 tabs, 4 figs, 30 refs.

References:

[1] 彭宏勤,张国伍.未来城市交通及其对未来城市发展影响[J].交通运输系统工程与信息,2020,20(1):2-5.
PENG Hong-qin, ZHANG Guo-wu. Future urban traffic and its influence on cities development[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(1): 2-5.(in Chinese)
[2] LIU Yang, LYU Cheng, LIU Zhi-yuan, et al. Exploring a large-scale multi-modal transportation recommendation system[J]. Transportation Research Part C: Emerging Technologies, 2021, 126: 103070.
[3] DAS D, SAHOO L, DATTA S. A survey on recommendation system[J]. International Journal of Computer Applications, 2017, 160(7): 6-10.
[4] ZHANG Qian, LU Jie, JIN Yao-chu. Artificial intelligence in recommender systems[J]. Complex and Intelligent Systems, 2021, 7(1): 439-457.
[5] CUI Ge, LUO Jun, WANG Xin. Personalized travel route recommendation using collaborative filtering based on GPS trajectories[J]. International Journal of Digital Earth, 2018, 11(3): 284-307.
[6] WEN Y T, YEO J Y, PENG W C, et al. Efficient keyword-aware representative travel route recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2017, 29(8): 1639-1652.
[7] 孙全明,常 磊,马 铖,等.基于图嵌入和CaGBDT的多模态出行推荐[J].北京邮电大学学报,2021,44(5):81-87,106.
SUN Quan-ming, CHANG Lei, MA Cheng, et al. Multi-modal transportation recommendation based on graph embedding and CaGBDT[J]. Journal of Beijing University of Posts and Telecommunications, 2021, 44(5): 81-87, 106.(in Chinese)
[8] 杨 敏,李宏伟,任怡凤,等.基于旅客异质性画像的公铁联程出行方案推荐方法[J].清华大学学报(自然科学版),2022,62(7):1220-1227.
YANG Min, LI Hong-wei, REN Yi-feng, et al. Road-rail intermodal travel recommendations based on a passenger heterogeneity profile[J]. Journal of Tsinghua University(Science and Technology), 2022, 62(7): 1220-1227.(in Chinese)
[9] SONG Xiang, DANAF M, ATASOY B, et al. Personalized menu optimization with preference updater: a Boston case study[J]. Transportation Research Record, 2018, 2672(8): 599-607.
[10] BALBONTIN C, HENSHER D A, COLLINS A T. How to better represent preferences in choice models: the contributions to preference heterogeneity attributable to the presence of process heterogeneity[J]. Transportation Research Part B: Methodological, 2019, 122: 218-248.
[11] GODDARD M. The EU general data protection regulation(GDPR): European regulation that has a global impact[J]. International Journal of Market Research, 2017, 59(6): 703-705.
[12] YANG Qiang, LIU Yang, CHEN Tian-jian, et al. Federated machine learning: concept and applications[J]. ACM Transactions on Intelligent Systems and Technology, 2019, 10(2): 1-19.
[13] JIAN Wei-tao, CHEN Kun-xu, HE Jun-shu, et al. A federated personal mobility service in autonomous transportation systems[J]. Mathematics, 2023, 11(12): 2693.
[14] LIU Sheng, CHEN Qi-yang, YOU Lin-lin. Fed2A: federated learning mechanism in asynchronous and adaptive modes[J]. Electronics, 2022, 11(9): 1393.
[15] WANG Si-hua, CHEN Ming-zhe, YIN Chang-chuan, et al. Federated learning for task and resource allocation in wireless high-altitude balloon networks[J]. IEEE Internet of Things Journal, 2021, 8(24): 17460-17475.
[16] CHENG Ke-wei, FAN Tao, JIN Yi-lun, et al. Secureboost: a lossless federated learning framework[J]. IEEE Intelligent Systems, 2021, 36(6): 87-98.
[17] ZHU Hang-yu, JIN Yao-chu. Multi-objective evolutionary federated learning[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(4): 1310-1322.
[18] MCMAHAN B, MOORE E, RAMAGE D, et al. Communication-efficient learning of deep networks from decentralized data[C]∥SINGH A, ZHU J. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics. Brookline: Microtome Publishing, 2017: 1273-1282.
[19] GELFAND A E. Gibbs sampling[J]. Journal of the American Statistical Association, 2000, 95(452): 1300-1304.
[20] MCFADDEN D. The measurement of urban travel demand[J]. Journal of Public Economics, 1974, 3(4): 303-328.
[21] HENSHER D A, GREENE W H. The mixed logit model: the state of practice[J]. Transportation, 2003, 30: 133-176.
[22] SONG Xiang. Personalization of future urban mobility[D]. Cambridge: Massachusetts Institute of Technology, 2018.
[23] MCFADDEN D, TRAIN K. Mixed MNL models for discrete response[J]. Journal of Applied Econometrics, 2000, 15(5): 447-470.
[24] TRAIN K. Mixed logit with a flexible mixing distribution[J]. Journal of Choice Modelling, 2016, 19: 40-53.
[25] LEE L F. Simulated maximum likelihood estimation of dynamic discrete choice statistical models some Monte Carlo results[J]. Journal of Econometrics, 1997, 82(1): 1-35.
[26] BANSAL P, KRUEGER R, BIERLAIRE M, et al. Bayesian estimation of mixed multinomial logit models: advances and simulation-based evaluations[J]. Transportation Research Part B: Methodological, 2020, 131: 124-142.
[27] CHIB S, GREENBERG E. Understanding the metropolis-Hastings algorithm[J]. The American Statistician, 1995, 49(4): 327-335.
[28] DANAF M, BECKER F, SONG Xiang, et al. Online discrete choice models: applications in personalized recommendations[J]. Decision Support Systems, 2019, 119: 35-45.
[29] BEN-AKIVA M, MCFADDEN D, TRAIN K. Foundations of stated preference elicitation: consumer behavior and choice-based conjoint analysis[J]. Foundations and Trends in Econometrics, 2019, 10(1/2): 1-144.
[30] DINH C T, TRAN N H, NGUYEN M N, et al. Federated learning over wireless networks: convergence analysis and resource allocation[J]. IEEE/ACM Transactions on Networking, 2020, 29(1): 398-409.

Memo

Memo:
-
Last Update: 2023-11-10