This repository is for the paper titled "Robust Subgraph Learning by Monitoring Early Training Representations" by Sepideh Neshatfar and Salimeh Yasaei Sekeh, published in IEEE IDS 2025, and won the Best Paper Award.
In order to find the subgraph indices for a specific dataset run the "Job.sh" shell available in the train folder.
To evaluate the indices of the resulting subgraph for different compression ratios, and Robustness-Performance combinations, run the "Job_Eval.sh" shell available in the Evaluate folder.