Bayesian Negative Sampling for Recommendation
Authors:
- 1 April 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 749-761
Abstract
How to sample high quality negative instances from unlabeled data, i.e., negative sampling, is important for training implicit collaborative filtering and contrastive learning models. Although previous studies have proposed some approaches to sample informative instances, discriminating false negative from true negative for unbiased negative sampling remains an unsolved problem. On the basis of our order relation analysis of negatives’ scores, we first derive the class conditional density of true negatives and that of false negatives. We next design a Bayesian classifier for negative classification, from which we define a model-agnostic posterior probability estimate of an instance being true negative as a quantitative negative signal measure. We also propose a Bayesian optimal sampling rule to sample high-quality negatives. The proposed Bayesian Negative Sampling (BNS) algorithm has a linear time complexity. Experimental studies validate the superiority of BNS over the peers in terms of better sampling quality and better recommendation performance. 1Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China
This publication has 38 references indexed in Scilit:
- Noise-Contrastive Estimation for Answer Selection with Deep Neural Networks2016
- Fast Matrix Factorization for Online Recommendation with Implicit Feedback2016
- Improving Latent Factor Models via Personalized Feature Projection for One Class Recommendation2015
- Context- and Content-aware Embeddings for Query Rewriting in Sponsored Search2015
- LINE2015
- Leveraging Social Connections to Improve Personalized Ranking for Collaborative Filtering2014
- ACRec2014
- Improving pairwise learning for item recommendation from implicit feedback2014
- Optimizing top-n collaborative filtering via dynamic negative item sampling2013
- Optimising area under the ROC curve using gradient descent2004