Deep Learning for Short-Term Traffic Flow Prediction

TL;DR

This paper presents a deep learning architecture that effectively predicts short-term traffic flows by capturing nonlinear spatio-temporal effects, demonstrated on real sensor data during events with sudden traffic regime changes.

Contribution

It introduces a novel deep learning model combining linear regularized and nonlinear layers specifically for traffic flow prediction, addressing sharp nonlinearities in traffic regimes.

Findings

Accurately predicts traffic flows during sudden regime changes.

Captures nonlinear spatio-temporal relations effectively.

Demonstrates superior performance on real-world sensor data.

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 $\ell_1$ 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.