DIPSGNN

This is our implementation for "Towards Differential Privacy in Sequential Recommendation: A Noisy Graph Neural Network Approach""

How to run.

1. Download dataset:

• ML-1M
• Yelp
• Tmall

2. Preprocess dataset:

cd datasets
python preprocess.py

3. run DIPSGNN with default hyperparameters:

python main_dipsgnn.py --dataset "ml-1m" --clip_norm 0.5 --epsilon1 20 --epsilon 4 --delta 2.5e-07 --step 1

Requirements

python: 3.8
torch: 1.12.1

Code Reference

• SRGNN