Learning-to-Rank at the Speed of Sampling: Plackett-Luce Gradient Estimation With Minimal Computational Complexity.
Harrie OosterhuisPublished in: CoRR (2022)
Keyphrases
- learning to rank
- gradient estimation
- computational complexity
- variance reduction
- ranking functions
- active sampling
- information retrieval
- loss function
- evaluation measures
- ranking svm
- collaborative filtering
- query dependent
- monte carlo
- document retrieval
- sample size
- direct optimization
- retrieval systems
- decision problems
- np hard
- supervised learning
- test collection
- random sampling
- learning to rank algorithms
- language model
- ranking models
- web search
- pairwise
- support vector
- decision trees