frankdoge / DGNAS-EP

DGNAS-EP is an automatical decoupled graph neural network (DGNN) designing method

Decoupled Graph Neural Architecture Search with Explainable Variable Propagation Operation

This repository contains the code implementation of Decoupled Graph Neural Architecture Search with Explainable Variable Propagation Operation submitted to Knowledge and Information Systems！

Running the Experiment

1. Searching the optimal DGNN Architecture

   To search for the optimal DGNN architecture, run the following command:

   python search-test.py
   

2. Test the accuracy of optimal DGNN Architecture

   To test the accuracy of optimal DGNN Architecture:

   python run-test.py