Bayesian committee of neural networks to predict travel times with confidence intervals

Short-term prediction of travel time is one of the central topics in current transportation research and practice. Among the more successful travel time prediction approaches are neural networks and combined prediction models (a 'committee'). However, both approaches have disadvantages. Usually many candidate neural networks are trained and the best performing one is selected. However, it is difficult and arbitrary to select the optimal network. In committee approaches a principled and mathematically sound framework to combine travel time predictions is lacking. This paper overcomes the drawbacks of both approaches by combining neural networks in a committee using Bayesian inference theory. An 'evidence' factor can be calculated for each model, which can be used as a stopping criterion during training, and as a tool to select and combine different neural networks. Along with higher prediction accuracy, this approach allows for accurate estimation of confidence intervals for the predictions. When comparing the committee predictions to single neural network predictions on the A12 motorway in the Netherlands it is concluded that the approach indeed leads to improved travel time prediction accuracy.

[1]  Nancy L. Nihan,et al.  Use of the box and Jenkins time series technique in traffic forecasting , 1980 .

[2]  J. W. C. van Lint,et al.  Short Term Traffic Prediction Models , 2007 .

[3]  H. J. Van Zuylen,et al.  Accurate freeway travel time prediction with state-space neural networks under missing data , 2005 .

[4]  J. W. C. van Lint,et al.  Online Learning Solutions for Freeway Travel Time Prediction , 2008, IEEE Transactions on Intelligent Transportation Systems.

[5]  Paulo J. G. Lisboa,et al.  A Bayesian neural network approach for modelling censored data with an application to prognosis after surgery for breast cancer , 2003, Artif. Intell. Medicine.

[6]  Hans van Lint,et al.  Reliable travel time prediction for freeways , 2004 .

[7]  Jiann-Shiou Yang.  Travel time prediction using the GPS test vehicle and Kalman filtering techniques , 2005, Proceedings of the 2005, American Control Conference, 2005..

[8]  H. M. Zhang,et al.  RECURSIVE PREDICTION OF TRAFFIC CONDITIONS WITH NEURAL NETWORK MODELS , 2000 .

[9]  Mark Dougherty,et al.  SHORT TERM INTER-URBAN TRAFFIC FORECASTS USING NEURAL NETWORKS , 1997 .

[10]  Der-Horng Lee,et al.  Short-term freeway traffic flow prediction : Bayesian combined neural network approach , 2006 .

[11]  Richard J. Hanowski,et al.  Driver Acceptance of Unreliable Traffic Information in Familiar and Unfamiliar Settings , 1997, Hum. Factors.

[12]  Stephen F. Gull,et al.  Developments in Maximum Entropy Data Analysis , 1989 .

[13]  Chris Bishop,et al.  Exact Calculation of the Hessian Matrix for the Multilayer Perceptron , 1992, Neural Computation.

[14]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[15]  Athanasios Kehagias,et al.  A Bayesian Multiple Models Combination Method for Time Series Prediction , 2001, J. Intell. Robotic Syst..

[16]  C.D. Mark,et al.  Predicting experienced travel time with neural networks: a PARAMICS simulation study , 2004, Proceedings. The 7th International IEEE Conference on Intelligent Transportation Systems (IEEE Cat. No.04TH8749).

[17]  J. W. C. van Lint,et al.  Bayesian Combination of Travel Time Prediction Models , 2008 .

[18]  Hojjat Adeli,et al.  Neural network model for rapid forecasting of freeway link travel time , 2003 .

[19]  Hecht-Nielsen.  Theory of the backpropagation neural network , 1989 .

[20]  Stephen D. Clark,et al.  Traffic Prediction Using Multivariate Nonparametric Regression , 2003 .

[21]  David J. C. MacKay,et al.  Bayesian Interpolation , 1992, Neural Computation.

[22]  David J. C. MacKay,et al.  A Practical Bayesian Framework for Backpropagation Networks , 1992, Neural Computation.

[23]  Anders Krogh,et al.  A Simple Weight Decay Can Improve Generalization , 1991, NIPS.

[24]  Ming Zhong,et al.  Refining Genetically Designed Models for Improved Traffic Prediction on Rural Roads , 2005 .

[25]  I Okutani,et al.  Dynamic prediction of traffic volume through Kalman Filtering , 1984 .

[26]  Martin Fodslette Møller,et al.  A scaled conjugate gradient algorithm for fast supervised learning , 1993, Neural Networks.

[27]  Steven I-Jy Chien,et al.  Development of a Hybrid Model for Dynamic Travel-Time Prediction , 2003 .

[28]  D. Nikovski,et al.  Univariate short-term prediction of road travel times , 2005, Proceedings. 2005 IEEE Intelligent Transportation Systems, 2005..

[29]  Anthony T. C. Goh,et al.  A hybrid Bayesian back‐propagation neural network approach to multivariate modelling , 2003 .

[30]  Michael J Demetsky,et al.  Multiple-Interval Freeway Traffic Flow Forecasting , 1996 .

[31]  Bart Baesens,et al.  Bayesian neural network learning for repeat purchase modelling in direct marketing , 2002, Eur. J. Oper. Res..

[32]  W. Press,et al.  Numerical Recipes: The Art of Scientific Computing , 1987 .

[33]  Satu Innamaa,et al.  Short-Term Prediction of Travel Time using Neural Networks on an Interurban Highway , 2005 .

[34]  Geoffrey E. Hinton,et al.  Learning internal representations by error propagation , 1986 .

[35]  Farid U. Dowla,et al.  Backpropagation Learning for Multilayer Feed-Forward Neural Networks Using the Conjugate Gradient Method , 1991, Int. J. Neural Syst..

[36]  David Mackay,et al.  Probable networks and plausible predictions - a review of practical Bayesian methods for supervised neural networks , 1995 .

[37]  Daniel B. Fambro,et al.  Application of Subset Autoregressive Integrated Moving Average Model for Short-Term Freeway Traffic Volume Forecasting , 1999 .