SIGNATURE WORK
CONFERENCE & EXHIBITION 2022

To achieve efficient management and control of the transportation system for improving its traffic efficiency and traffic safety, Intelligent Transportation Systems (ITS) have been introduced and received a lot of attention from academia and industry in recent years. Serving as an essential component of ITS, traffic flow prediction (TFP) has gained more thoughtful attention because it can provide predictive and timely information for the whole society.

However, some of the well-developed algorithms still have some limitations. Usually, researchers tend to view TFP as an academic problem; therefore, prediction accuracy might be their primary evaluation metric. However, if we treat TFP as an application problem, time efficiency is equally important besides accuracy.

The main contributions of this paper are as follows:

1. Quantify the relationship between the time cost of some given algorithms and the size of the road network.
2. Applied Auto-Encoder(AE) as a dimension reduction technique to shorten the training time of the extensive road network.