Deep Architecture for Traffic Flow Prediction: Deep Belief Networks With Multitask Learning

被引:850
作者
Huang, Wenhao [1 ]
Song, Guojie [1 ,2 ]
Hong, Haikun [1 ]
Xie, Kunqing [1 ,3 ]
机构
[1] Peking Univ, Sch Elect Engn & Comp Sci, Beijing 100871, Peoples R China
[2] Peking Univ, Res Ctr Intelligent Informat Proc, Beijing 100871, Peoples R China
[3] Peking Univ, Res Ctr Intelligent Traff Syst, Beijing 100871, Peoples R China
基金
国家高技术研究发展计划(863计划);
关键词
Deep learning; multitask learning (MTL); task grouping; traffic flow prediction; INTELLIGENT TRANSPORTATION SYSTEMS;
D O I
10.1109/TITS.2014.2311123
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Traffic flow prediction is a fundamental problem in transportation modeling and management. Many existing approaches fail to provide favorable results due to being: 1) shallow in architecture; 2) hand engineered in features; and 3) separate in learning. In this paper we propose a deep architecture that consists of two parts, i.e., a deep belief network (DBN) at the bottom and a multitask regression layer at the top. A DBN is employed here for unsupervised feature learning. It can learn effective features for traffic flow prediction in an unsupervised fashion, which has been examined and found to be effective for many areas such as image and audio classification. To the best of our knowledge, this is the first paper that applies the deep learning approach to transportation research. To incorporate multitask learning (MTL) in our deep architecture, a multitask regression layer is used above the DBN for supervised prediction. We further investigate homogeneous MTL and heterogeneous MTL for traffic flow prediction. To take full advantage of weight sharing in our deep architecture, we propose a grouping method based on the weights in the top layer to make MTL more effective. Experiments on transportation data sets show good performance of our deep architecture. Abundant experiments show that our approach achieved close to 5% improvements over the state of the art. It is also presented that MTL can improve the generalization performance of shared tasks. These positive results demonstrate that deep learning and MTL are promising in transportation research.
引用
收藏
页码:2191 / 2201
页数:11
相关论文
共 37 条
[1]  
[Anonymous], 1998, MULTITASK LEARNING
[2]  
[Anonymous], 2008, P 25 INT C MACH LEAR, DOI DOI 10.1145/1390156.1390177
[3]  
[Anonymous], 2008, Advances in neural information processing systems
[4]  
[Anonymous], 2000, HIGHW CAP MAN
[5]   Online-SVR for short-term traffic flow prediction under typical and atypical traffic conditions [J].
Castro-Neto, Manoel ;
Jeong, Young-Seon ;
Jeong, Myong-Kee ;
Han, Lee D. .
EXPERT SYSTEMS WITH APPLICATIONS, 2009, 36 (03) :6164-6173
[6]   Neural-Network-Based Models for Short-Term Traffic Flow Forecasting Using a Hybrid Exponential Smoothing and Levenberg-Marquardt Algorithm [J].
Chan, Kit Yan ;
Dillon, Tharam S. ;
Singh, Jaipal ;
Chang, Elizabeth .
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, 2012, 13 (02) :644-654
[8]  
Dean J., 2012, Advances in Neural Information Processing Systems, P1223, DOI DOI 10.5555/2999134.2999271
[9]  
Deng L, 2012, INT CONF ACOUST SPEE, P2133, DOI 10.1109/ICASSP.2012.6288333
[10]   Bayesian time-series model for short-term traffic flow forecasting [J].
Ghosh, Bidisha ;
Basu, Biswajit ;
O'Mahony, Margaret .
JOURNAL OF TRANSPORTATION ENGINEERING, 2007, 133 (03) :180-189