Skip to content

Latest commit

 

History

History
executable file
·
51 lines (42 loc) · 2.33 KB

README.md

File metadata and controls

executable file
·
51 lines (42 loc) · 2.33 KB
Raw

Spatio-Temporal Graph Convolutional Networks

issues forks stars License

About

The PyTorch version of STGCN implemented for the paper Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting.

Paper

https://arxiv.org/abs/1709.04875

Related works

  1. TCN: An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
  2. GLU and GTU: Language Modeling with Gated Convolutional Networks
  3. ChebNet: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering
  4. GCN: Semi-Supervised Classification with Graph Convolutional Networks

Related code

  1. TCN: https://github.com/locuslab/TCN
  2. ChebNet: https://github.com/mdeff/cnn_graph
  3. GCN: https://github.com/tkipf/pygcn

Dataset

Source

  1. METR-LA: DCRNN author's Google Drive
  2. PEMS-BAY: DCRNN author's Google Drive
  3. PeMSD7(M): STGCN author's GitHub repository

Preprocessing

Using the formula from ChebNet：

Model structure

Differents of code between mine and author's

  1. Fix bugs
  2. Add Early Stopping approach
  3. Add Dropout approach
  4. Offer a different set of hyperparameters
  5. Offer config files for two different categories graph convolution (ChebyGraphConv and GraphConv)
  6. Add datasets METR-LA and PEMS-BAY
  7. Adopt a different data preprocessing method

Requirements

To install requirements:

pip3 install -r requirements.txt