This is the pytorch implementation of our paper: Unified Denoising Training for Recommendation.
Haoyan Chua, Yingpeng Du, Zhu Sun, Ziyan Wang and Jie Zhang.
- python 3.11.5
- pytorch 2.0.1
- numpy 1.26.3
We provided the code for training and inference for GMF, NeuMF & CDAE:
Go to NCF folder and simply run the code below with default settings to return results shown in the paper:
python main.py --model GMF --dataset amazon_book --gpu 0
or for Neumf
python main.py --model NeuMF-end --dataset movielens --gpu 0
Go to CDAE folder and simply run the code below with default settings to return results shown in the paper:
python main_CDAE.py --dataset yelp --gpu 0
To change the hyperparameter settings, --userfact1 & --userfact2 controls [a , b] in the paper while --temp1 & --temp2 controls [a' , b'].
To save training time and start validation evaluation later, --epoch_eval determines the epoch to begin evaluation.
- Train GMF on Yelp with different hyperparameter settings:
python main.py --dataset yelp --model GMF --userfact1 0.05 --userfact2 0.0 --temp1 0.1 --temp2 0.5 --epoch_eval 30 --gpu=0
- Train CDAE on Amazon-book with different hyperparameter settings:
python main_CDAE.py --dataset amazon_book --model CDAE --userfact1 1.0 --userfact2 0.5 --temp1 0.5 --temp2 1.0 --epoch_eval 30 --gpu=0
If you use our code, please kindly cite: