Abstract
Over the past decade, many approaches have been introduced for traffic speed prediction. However, providing fine-grained, accurate, time-efficient, and adaptive traffic speed prediction for a growing transportation network where the size of the network keeps increasing and new traffic detectors are constantly deployed has not been well studied. To address this issue, this paper presents DistTune based on long short-term memory (LSTM) and the Nelder-Mead method. When encountering an unprocessed detector, DistTune decides if it should customize an LSTM model for this detector by comparing the detector with other processed detectors in the normalized traffic speed patterns they have observed. If a similarity is found, DistTune directly shares an existing LSTM model with this detector to achieve time-efficient processing. Otherwise, DistTune customizes an LSTM model for the detector to achieve fine-grained prediction. To make DistTune even more time-efficient, DisTune performs on a cluster of computing nodes in parallel. To achieve adaptive traffic speed prediction, DistTune also provides LSTM re-customization for detectors that suffer from unsatisfactory prediction accuracy due to, for instance, changes in traffic speed patterns. Extensive experiments based on traffic data collected from freeway I5-N in California are conducted to evaluate the performance of DistTune. The results demonstrate that DistTune provides fine-grained, accurate, time-efficient, and adaptive traffic speed prediction for a growing transportation network.
Highlights
Traffic speed is a key indicator to measure the efficiency of a transportation network
We focus on evaluating Bayesian optimization approach (BOA) and Nelder-Mead method (NMM) to see which of them suits DistTune better, i.e., which of them can more time efficiently find Long Short-Term Memory (LSTM) hyperparameters for detectors such that the resulting LSTM models are able to achieve the desired prediction accuracy
We have introduced DistTune, a distributed scheme to achieve fine-grained, accurate, time-efficient, and adaptive traffic speed prediction for the increasing number of detectors deployed in a growing transportation network
Summary
Traffic speed is a key indicator to measure the efficiency of a transportation network. Many approaches and methods have been introduced for traffic speed prediction. Each of them can be summarized as learning a mapping function between input variables and output variables. These methods can be classified into two main types: parametric and nonparametric. Parametric approaches simplify the mapping function to a known form, i.e., they require a predefined model. The forget gate determines if current memory contents should be deleted. The output gate decides if current memory contents should be output. This design enables LSTM to preserve information over long time lags, addressing the vanishing gradient problem [9]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Transportation Research Record: Journal of the Transportation Research Board
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
