Understanding Dynamic Spatio-Temporal Contexts in Long Short-Term Memory for Road Traffic Speed Prediction

TL;DR

This paper introduces a localized LSTM model that captures dynamic spatial and temporal dependencies for more accurate road traffic speed prediction, outperforming baseline methods.

Contribution

The study presents a novel localized LSTM with dynamic spatial weights to model complex interactions in traffic data, enhancing prediction accuracy.

Findings

Superior prediction performance over baseline methods

Effective modeling of dynamic spatial-temporal dependencies

Handling long-term dependencies and non-linear features

Abstract

Reliable traffic flow prediction is crucial to creating intelligent transportation systems. Many big-data-based prediction approaches have been developed but they do not reflect complicated dynamic interactions between roads considering time and location. In this study, we propose a dynamically localised long short-term memory (LSTM) model that involves both spatial and temporal dependence between roads. To do so, we use a localised dynamic spatial weight matrix along with its dynamic variation. Moreover, the LSTM model can deal with sequential data with long dependency as well as complex non-linear features. Empirical results indicated superior prediction performances of the proposed model compared to two different baseline methods.