Hard Negatives or False Negatives: Correcting Pooling Bias in Training Neural Ranking Models.
Yinqiong CaiJiafeng GuoYixing FanQingyao AiRuqing ZhangXueqi ChengPublished in: CoRR (2022)
Keyphrases
- false negative
- false positives
- false negative rate
- low false positive rate
- ranking models
- learning to rank
- false positive rate
- detection rate
- ranking functions
- training set
- web search
- ranking algorithm
- learning to rank algorithms
- data sets
- precision and recall
- benchmark datasets
- detection algorithm
- collaborative filtering
- semi supervised
- computer vision