臺灣博碩士論文加值系統

交通安全上的問題一直都是很重要的議題，絕大部分的交通事故都是因為過於激進的駕駛行為。對於保險公司而言，如果能夠使用駕駛者的駕駛行為來判斷給予的保險金額是比較適合的，而不是依據車種或駕駛者的形象來給予保險金，為了要對駕駛者的行為做定義，我們必須能夠找出這些行車路徑分別是屬於哪些駕駛者。

我們會針對兩件事情進行討論及分析，第一個是從序列資料中進行學習並產生相對應合適的特徵藉由字典學習及稀疏矩陣。第二個是當我們使用稀疏矩陣來產生特徵時，我們可以使用二元分類來將不同行為的駕駛者分出來，並且使用全資料模型、時間模型及地區模型來進行分析，最後我們可以找出最影響駕駛者行為的模型來將駕駛者並辨識出行車路徑屬於哪位駕駛者。

Traffic safety has been an important issue for a long time. Most of the traffic accident are happened because of the aggressive driving behavior. For insurance company, understanding the driver's behavior is more appropriate to define the insurance instead of making decisions based on vehicle and driver characteristics. To define the behavior for drivers, we need to classify the trip data for different driver.

In this research, we focus on two things. First one is learning the sequential data and generate appropriate features by dictionary learning and sparse coding. Second one is using the sparse code features with binary classification to identify the trips between drivers. We also analyze three models for driving behaviors: full-data model, time model and location model. We get the best model which affect the drivers' behavior more to identify the trip for drivers.

Abstract in Chinese . . . . . . . . . . . . . . . . . . . . . . . . . . i
Abstract in English . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Research Framework . . . . . . . . . . . . . . . . . . . . 3
1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Dictionary Learning and Sparse Coding . . . . . . . . . . 11
3.2 Dictionary Learning and Sparse Coding Analysis . . . . . 14
3.2.1 Dictionary Learning Analysis . . . . . . . . . . . 14
3.2.2 Sparse Coding Analysis . . . . . . . . . . . . . . 17
3.2.3 Parameter Selection . . . . . . . . . . . . . . . . 18
3.3 Support Vector Machine . . . . . . . . . . . . . . . . . . 21
3.3.1 RBF Kernel . . . . . . . . . . . . . . . . . . . . . 22
4 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.1 Driving Behavior Analysis . . . . . . . . . . . . . . . . . 24
4.1.1 Dataset . . . . . . . . . . . . . . . . . . . . . . . 24
4.2 Data Preprocessing . . . . . . . . . . . . . . . . . . . . . 26
4.2.1 Model . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2.2 Model Comparison . . . . . . . . . . . . . . . . . 35
4.2.3 Comparison the Parameter of Dictionary . . . . . 36
4.2.4 Kaggle Competition . . . . . . . . . . . . . . . . 38
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Letter of Authority . . . . . . . . . . . . . . . . . . . . . . . . . . 42

[1] J. Mairal, F. Bach, J. Ponce, and G. Sapiro, “Online dictionary learning for sparse coding,” in Proceedings
of the 26th annual international conference on machine learning, pp. 689–696, ACM, 2009.
[2] I. Tosic and P. Frossard, “Dictionary learning,” IEEE Signal Processing Magazine, vol. 28, no. 2,
pp. 27–38, 2011.
[3] J. Mairal, F. Bach, and J. Ponce, “Task-driven dictionary learning,” IEEE Transactions on Pattern
Analysis and Machine Intelligence, vol. 34, no. 4, pp. 791–804, 2012.
[4] D. Vainsencher, S. Mannor, and A. M. Bruckstein, “The sample complexity of dictionary learning,”
Journal of Machine Learning Research, vol. 12, no. Nov, pp. 3259–3281, 2011.
[5] J. Mairal, J. Ponce, G. Sapiro, A. Zisserman, and F. R. Bach, “Supervised dictionary learning,” in
Advances in neural information processing systems, pp. 1033–1040, 2009.
[6] H. Lee, A. Battle, R. Raina, and A. Y. Ng, “Efficient sparse coding algorithms,” in Advances in neural
information processing systems, pp. 801–808, 2006.
[7] D. L. Donoho and M. Elad, “Optimally sparse representation in general (nonorthogonal) dictionaries
via ℓ1 minimization,” Proceedings of the National Academy of Sciences, vol. 100, no. 5, pp. 2197–
2202, 2003.
[8] K. Huang and S. Aviyente, “Sparse representation for signal classification,” in Advances in neural
information processing systems, pp. 609–616, 2006.
[9] P. A. Ellison, J. M. Govern, H. L. Petri, and M. H. Figler, “Anonymity and aggressive driving behavior:
A field study,” Journal of Social Behavior and Personality, vol. 10, no. 1, p. 265, 1995.
[10] D. T. Kenrick and S. W. MacFarlane, “Ambient temperature and horn honking a field study of the heat/
aggression relationship,” Environment and behavior, vol. 18, no. 2, pp. 179–191, 1986.
[11] K. K. Deaux, “Honking at the intersection: A replication and extension,” The Journal of Social Psychology,
vol. 84, no. 1, pp. 159–160, 1971.
[12] C. W. Turner, J. F. Layton, and L. S. Simons, “Naturalistic studies of aggressive behavior: aggressive
stimuli, victim visibility, and horn honking.,” Journal of personality and social psychology, vol. 31,
no. 6, p. 1098, 1975.
[13] H. Eren, S. Makinist, E. Akin, and A. Yilmaz, “Estimating driving behavior by a smartphone,” in
Intelligent Vehicles Symposium (IV), 2012 IEEE, pp. 234–239, IEEE, 2012.
[14] Z. Sun and X. J. Ban, “Vehicle classification using gps data,” Transportation Research Part C: Emerging
Technologies, vol. 37, pp. 102–117, 2013.
[15] C. Cortes and V. Vapnik, “Support-vector networks,” Machine learning, vol. 20, no. 3, pp. 273–297,
1995.
[16] C.-W. Hsu, C.-C. Chang, C.-J. Lin, et al., “A practical guide to support vector classification,” 2003.
[17] S. P. M. Deployment and the members of the test conductor team, “Safety pilot model deployment–
sample data,” 2014.