ARSM Gradient Estimator for Supervised Learning to Rank
This work addresses ranking problems in information retrieval and recommendation systems, presenting an incremental improvement with flexibility in loss function choice.
The authors tackled the problem of supervised learning to rank by proposing a model where relevance labels follow a categorical distribution based on a scoring function, optimized with an unbiased low-variance gradient estimator. Their method achieved better or comparable results on two datasets compared to existing pairwise and listwise approaches.
We propose a new model for supervised learning to rank. In our model, the relevance labels are assumed to follow a categorical distribution whose probabilities are constructed based on a scoring function. We optimize the training objective with respect to the multivariate categorical variables with an unbiased and low-variance gradient estimator. Learning-to-rank methods can generally be categorized into pointwise, pairwise, and listwise approaches. Although our scoring function is pointwise, the proposed framework permits flexibility over the choice of the loss function. In our new model, the loss function need not be differentiable and can either be pointwise or listwise. Our proposed method achieves better or comparable results on two datasets compared with existing pairwise and listwise methods.