Deep learning for short-term traffic flow prediction

By Nicholas G. Polson, Vadim O. Sokolov. 2016

960 citations recorded by the source corpus.

Abstract We develop a deep learning model to predict traffic flows. The main contribution is development of an architecture that combines a linear model that is fitted using l 1 regularization and a sequence of tanh layers. The challenge of predicting traffic flows are the sharp nonlinearities due to transitions between free flow, breakdown, recovery and congestion. We show that deep learning architectures can capture these nonlinear spatio-temporal effects. The first layer identifies spatio-temporal relations among predictors and other layers model nonlinear relations. We illustrate our methodology on road sensor data from Interstate I-55 and predict traffic flows during two special events; a Chicago Bears football game and an extreme snowstorm event. Both cases have sharp traffic flow regime changes, occurring very suddenly, and we show how deep learning provides precise short term traffic flow predictions.

Subfields: Building and Construction, Control and Systems Engineering, Signal Processing

View Paper: https://www.semanticscholar.org/paper/566de8271556e8776a1b0cef86a045ef3a92d214

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Building and Construction

In Control and Systems Engineering

In Signal Processing

Citations in the corpus

Deep Spatial–Temporal 3D Convolutional Neural Networks for Traffic Data Forecasting (2019, methodology): “Compared with the above methods, data-driven methods provide more ﬂexible alternatives for trafﬁc forecasting [9].”
Flow: A Modular Learning Framework for Mixed Autonomy Traffic (2017, methodology): “Deep RL has been used for trafﬁc prediction [85], [86] and control [87], [88], [89].”
High-Order Graph Convolutional Recurrent Neural Network: A Deep Learning Framework for Network-Scale Traffic Learning and Forecasting (2018, background): “Deep learning models have shown their superior capabilities of capturing nonlinear spatiotemporal effects for traffic forecasting [29].”
Flow: Architecture and Benchmarking for Reinforcement Learning in Traffic Control (2017, methodology): “A deep RL architecture was used in [74] to predict traffic flows, demonstrating success even during special events with nonlinear features; to learn features to represent states”
Deep Learning for Spatio-Temporal Data Mining: A Survey (2019, background): “Application domains Related works Transportation [177], [187], [116], [31], [29], [6], [32], [127], [53], [112], [26], [105], [166], [82], [23], [66], [108], [126], [81], [139], [78], [128], [146], [148], [72], [98], [189], [22], [24], [99], [96], [16], [97], [85], [15], [164], [165], [95], [140], [”
Spatial temporal incidence dynamic graph neural networks for traffic flow forecasting (2020, methodology): “Recent years, deep learning methods have shown promising results in dynamic prediction over spatial-temporal traffic flows data [1, 12, 5].”
Short-Term Forecasting of Passenger Demand under On-Demand Ride Services: A Spatio-Temporal Deep Learning Approach (2017, background): “Polson and Sokolov (2017) found that the sharp nonlinearities of traffic flow, as a result of transitions between the free flow, breakdown, recovery and congestion, could be captured by a DL architecture.”
Urban traffic signal control with connected and automated vehicles: A survey (2019, neutral): “So far, most stochastic filtering and particle filter techniques are model-based and try to integrate traffic flow models with data; while the pure data-driven methods such as deep-learning based methods depend solely on data (Lv et al., 2015; Ma et al., 2015, 2017; Polson and Sokolov, 2017).”
Short-term traffic flow prediction based on spatio-temporal analysis and CNN deep learning (2019, background): “Reliable traffic prediction information not only provides a reference for the formulation of management plans and allocation of traffic resources, but also enables travelers to plan their journeys (Polson and Sokolov 2017).”
GeoAI: spatially explicit artificial intelligence techniques for geographic knowledge discovery and beyond (2019, neutral): “…from scanned historical maps (Duan et al . in this issue), semantic classi ﬁ cation (e.g. LiDAR point clouds) (Guo and Feng in this issue), novel methods for spatial interpolation (Zhu et al . in this issue), and advances in tra ﬃ c forecasting (Polson and Sokolov 2017, Ren et al . in this issue).”
Graph Neural Networks for Intelligent Transportation Systems: A Survey (2023, background): “…of traffic variables using emerging sensors have shifted traffic forecasting efforts towards data-driven methods, and numerous attempts have been made to develop flexible, large-scale, and real-time traffic forecasting models [22], [25], especially using deep learning methods [23], [26].”