MLGO

This is a PyTorch implementation of the paper: MLGO: Multi-Layer graph neural ODEs for traffic forecasting - PubMedpublished in Neural Networks-2026.

Requirements

The model is implemented using Python3 with dependencies as followed: tqdm==4.67.3

torch==2.11.0

torchdiffeq==0.2.5

fastdtw==0.3.4

numpy==2.4.4

Data Preparation

PeMS Datasets

Download data folder ,which contains PeMS03, PeMS04, PeMS07, PeMS08, and PeMS-BAY datasets from https://github.com/guoshnBJTU/ASTGNN. Uncompress them.

Model Training

PeMS03

python run.py --config ./configs/pems03.conf

PeMS04

python run.py --config ./configs/pems04.conf

PeMS07

python run.py --config ./configs/pems07.conf

PeMS08

python run.py --config ./configs/pems08.conf

PeMS-BAY

python run.py --config ./configs/pemsbay.conf

Citation

@article{gao2026connecting,
  title={MLGO: Multi-Layer graph neural ODEs for traffic forecasting},
  author={Gao, Mengzhou and Yu, Huangqian and Jiao, Pengfei},
  journal={Neural Networks},
  year={2026}
}