使用者的交通模式(例如：走路、公車、汽車)反映其戶外行為模式。隨著具GPS (Global Positioning System，全球衛星定位系統)功能的手機與行動上網的普及，即時預測交通模式成為下列三種應用的基礎：消費行為、旅行行程分享、智慧路徑推薦。預測交通模式是本篇論文的核心議題，我們使用一個兩層推論架構來處理。第一層以五種特徵推論更換點，也就是交通模式被改變地點。第二層以十種特徵推論交通模式，七種交通模式被考慮：走路、單車、公車、汽車、機車、捷運、火車。第一層的F-measure是0.753。第二層的實驗結果以兩種指標來評估：距離準確度(AL)與時間準確度(AD)。其中，距離準確度為 0.876，時間準確度為 0.693。我們的題目比相關文獻更具挑戰性，因為我們的交通模式更多，而更精細的分類是較困難的。

The transportation mode of users, such as Walk, Bus, or Car, indicates the outdoor behavior pattern of the user. As the GPS (Global Positioning System) enabled phones and mobile internet accesses become pervasive, the prediction of transportation mode becomes fundamental in the area of shopping behaviors, travel itinerary sharing and smart route recommendation. Learning transportation modes is the central issue and a two-level inference architecture is used. The first level learns change-points, locations whose transportation mode differs from the previous location, with five features. The second level learns seven transportation modes, Walk, Bike, Bus, Car, Moto (Motorcycle), MRT (Mass Rapid Transit), and Train, with ten features. The F-measure is 0.753 in the first level. The results of second level are evaluated by Accuracy by by Length (AL) and Accuracy by Duration (AD), respectively. AL = 0.876 and AD = 0.693. Comparing to the related works, which contains four to five modes at the most, our work is more challenging since we have seven modes and the fine-grained classification is more difficult. The two main challenges in the classification of transportation modes, change-points and traffic congestions, are adressed and the combination of more sensors with GPS, such as 3-axis accelerometer, could be the future improvements.

[1] Yu Zheng, Quannan Li, Yukun Chen, Xing Xie, Wei Y. Ma, “Understanding mobility based on GPS data,” In UbiComp ''08: Proceedings of the 10th international conference on Ubiquitous computing (2008), pp. 312-321.
[2] Yu Zheng , Like Liu , Longhao Wang , Xing Xie, “Learning Transportation Mode from Raw GPS Data for Geographic Applications on the Web,” Proceeding of the 17th international conference on World Wide Web, 2008.
[3] Sasank Reddy, Min Mun, Jeff Burke, Deborah Estrin, Mark Hansen, and Mani Srivastava. 2010. Using mobile phones to determine transportation modes. ACM Trans. Sen. Netw. 6, 2, Article 13 (March 2010), 27 pages.
[4] L. Liao, D. J. Patterson, D. Fox, and H. Kautz, “Learning and Inferring Transportation Routines,” in Artificial Intelligence, 2007.
[5] D. Patterson, L. Liao, D. Fox, and H. Kautz, “Inferring High-Level Behavior from Low-Level Sensors,” in Proceedings of The Fifth International Conference on Ubiquitous Computing (UBICOMP), 2003.
[6] A. Doucet, N. de Freitas, N. Gordon (Eds.), Sequential Monte Carlo in Practice, Springer-Verlag, New York, 2001.
[7] S. Thrun, W. Burgard, D. Fox, Probabilistic Robotics, MIT Press, 2005.
[8] J. Bilmes, “A gentle tutorial on the EM algorithm and its application to parameter estimation for Gaussian mixture and hidden Markov models,” in Technical Report ICSI-TR-97-021, University of Berkeley, 1998.
[9] L. R. Rabiner, “A tutorial on hidden Markov models and selected applications in speech recognition,” in Proceedings of the IEEE, 1989, IEEE Log Number 8825949.
[10] G. McLachlan, T. Krishnan, “The EM Algorithm and Extensions,” John Wiley & Sons, New York, 1997.
[11] T. Dean, K. Kanazawa, “Probabilistic temporal reasoning,” in: Proc. of the National Conference on Artificial Intelligence (AAAI), 1988.
[12] K. Murphy, “Dynamic Bayesian networks: Representation, inference and learning,” Ph.D. thesis, UC Berkeley, Computer Science Division, 2002.
[13] A. Doucet, J. de Freitas, K. Murphy, S. Russell, “Rao-Blackwellised particle filtering for dynamic Bayesian networks,” in: Proc. of the Conference on Uncertainty in Artificial Intelligence (UAI), 2000.
[14] J. Lafferty, A. McCallum, F. Pereira, “Conditional random fields: Probabilistic models for segmenting and labeling sequence data,” in Proc. ICML, 2001.
[15] L. Liao, D. J. Patterson, D. Fox, and H. Kautz, “Building Personal Maps from GPS Data,” in New York Academy of Sciences, 2007.
[16] L. Liao, D. Fox, H. Kautz, “Location-based activity recognition,” in Advances in Neural Information Processing Systems (NIPS), 2005.
[17] L. Liao, D. Fox, H. Kautz, “Hierarchical conditional random fields for GPS-based activity recognition,” in Robotics Research: The Twelfth International Symposium (ISRR-05), Springer Verlag, 2006.
[18] L. Liao, D. Fox, and H. Kautz. “Extracting Places and Activities from GPS Traces Using Hierarchical Conditional Random Fields,” in International Journal of Robotics Research, 2007.
[19] A. Anagnostopoulos, M. Vlachos, M. Hadjieleftheriou, E. Keogh, P.S. Yu, “Global Distance-Based Segmentation of Trajectories,” in KDD, pp. 34-43, 2006.
[20] S. Elnekave, M. Last, O. Maimon, B.S. Yehuda, H. Einsiedler, M. Friedman, M. Siebert, “Discovering Regular Groups of Mobile Objects Using Incremental Clustering,” in WPNC, 2008.
[21] C.R. Jose, H. Tang and I. Espigares, “Consistent LBS Solution in Next Generations of Mobile Internet”, Parallel and Distributed Systems. Proceedings. Ninth International Conference on 17-20 Page(s):637 – 642, 2002
[22] Elnekave, S., Last, M., and Maimon, O. 2007. Predicting future locations using clusters'' centroids. In Proceedings of the 15th Annual ACM international Symposium on Advances in Geographic information Systems (Seattle, Washington, November 07 - 09, 2007). GIS ''07.
[23] Xiao Hu, Danil V. Prokhorov, Donald C. Wunsch II, Time series prediction with a weighted bidirectional multi-stream extended Kalman filter, Neurocomputing, Volume 70, Issues 13-15, Selected papers from the 3rd International Conference on Development and Learning (ICDL 2004), 3rd International Conference on Development and Learning; Time series prediction competition: the CATS benchmark, August 2007 Gang Chen, Fei Wang, and Changshui Zhang. (2007). Collaborative filtering using orthogonal nonnegative matrix tri-factorization. Inf. Process. Manage, Vol. 45, Pergamon Press, Inc., Tarrytown, NY, USA, (pp. 368-379).
[24] Four Out of Five Cell Phones to Integrate GPS by End of 2011
http://www.isuppli.com/Mobile-and-Wireless-Communications/News/Pages/Four-out-of-Five-Cell-Phones-to-Integrate-GPS-by-End-of-2011.aspx
[25] Google Map http://maps.google.com/
[26] Anderson, I., and Muller, H. 2006. Practical activity recognition using GSM data. Technical Report CSTR-06-016, Department of Computer Science, University of Bristol.
[27] Miluzzo, E., Lane, N., Fodor, K., Peterson, R., Lu, H., Musolesi, M., Eisenman, S., Zheng, X., and Campbell, A. 2008. Sensing meets mobile social networks: The design implementation and evaluation of the cenceMe application. In Proceedings of the 6th ACM Conference on Embedded Network Sensor Systems. ACM, New York. 337–350.
[28] Mun, M., Estrin, D., Burke, J., and Hansen, M. 2008. Parsimonious mobility classification using GSM and WiFi traces. In Proceedings of the 5th Workshop on Embedded Networked Sensors (HotEmNets).
[29] Sohn, T., Varshavsky, A., LaMarca, A., Chen, M., Choudhury, T., Smith, I., Consolvo, S., Hightower, J., Griswold, W., and deLara, E. 2006. Mobility detection using everyday GSM traces. In Lecture Notes in Computer Science, vol. 4206, Springer-Verlag, Berlin, Germany, 212.
[30] P.A. Gonzalez, J.S. Weinstein, S.J. Barbeau, M.A. Labrador, P.L. Winters, N.L. Georggi, and R. Perez. 2010. Automating mode detection for travel behaviour analysis by using global positioning systems-enabled mobile phones and neural networks. IET Intell. Transp. Syst. 4, 37