[TITS2025] Pattern-Matching Dynamic Memory Network for Dual-Mode Traffic Prediction

This is a PyTorch implementation of Pattern-Matching Dynamic Memory Network for Dual-Mode Traffic Prediction

Update

(2025/4/22)

• Good news! This paper is accepted by IEEE Transactions on Intelligent Transportation Systems.

Table of Contents

• configs: training Configs and model configs for each dataset

• lib: contains self-defined modules for our work, such as data loading, data pre-process, normalization, and evaluate metrics.

• model: implementation of our model

• data: contains relevant datasets

Requirements

Python 3.6.5, Pytorch 1.9.0, Numpy 1.16.3, argparse and configparser

Data Preparation

For convenience, we package these datasets used in our model in Google Drive.

Unzip the downloaded dataset files into the data folder.

Model Training

python run.py --datasets {DATASET_NAME} --type {MODEL_TYPE} --mode {MODE_NAME} 

Replace {DATASET_NAME} with one of datasets.

such as python run.py --dataset NYC-Taxi16

To run PM-DMNet with the desired configuration, set the type parameter accordingly:

• Set type P to run PM-DMNet(P).
• Set type R to run PM-DMNet(R).

There are two options for {MODE_NAME} : train and test

Selecting train will retrain the model and save the trained model parameters and records in the experiment folder.

With test selected, run.py will import the trained model parameters from {DATASET_NAME}.pth in the 'pre-trained' folder.

Here is an example of how to run the script using the specified parameters:

python run.py --dataset PEMSD8 --type P --mode train

Cite

If you find the paper useful, please cite as following:

@article{weng2025pattern,
  title={Pattern-Matching Dynamic Memory Network for Dual-Mode Traffic Prediction},
  author={Weng, Wenchao and Wu, Mei and Jiang, Hanyu and Kong, Wanzeng and Kong, Xiangjie and Xia, Feng},
  journal={IEEE Transactions on Intelligent Transportation Systems},
  year={2025},
  publisher={IEEE}
}

More Related Works

• [Neural Networks] RGDAN: A random graph diffusion attention network for traffic prediction

• [Pattern Recognition] A Decomposition Dynamic Graph Convolutional Recurrent Network for Traffic Forecasting

• [Neural Networks] PDG2Seq: Periodic Dynamic Graph to Sequence model for Traffic Flow Prediction