Which Tricks are Important for Learning to Rank?

Ivan Lyzhin, Aleksei Ustimenko, Andrey Gulin, Liudmila Prokhorenkova
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:23264-23278, 2023.

Abstract

Nowadays, state-of-the-art learning-to-rank methods are based on gradient-boosted decision trees (GBDT). The most well-known algorithm is LambdaMART which was proposed more than a decade ago. Recently, several other GBDT-based ranking algorithms were proposed. In this paper, we thoroughly analyze these methods in a unified setup. In particular, we address the following questions. Is direct optimization of a smoothed ranking loss preferable over optimizing a convex surrogate? How to properly construct and smooth surrogate ranking losses? To address these questions, we compare LambdaMART with YetiRank and StochasticRank methods and their modifications. We also propose a simple improvement of the YetiRank approach that allows for optimizing specific ranking loss functions. As a result, we gain insights into learning-to-rank techniques and obtain a new state-of-the-art algorithm.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-lyzhin23a, title = {Which Tricks are Important for Learning to Rank?}, author = {Lyzhin, Ivan and Ustimenko, Aleksei and Gulin, Andrey and Prokhorenkova, Liudmila}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {23264--23278}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/lyzhin23a/lyzhin23a.pdf}, url = {https://proceedings.mlr.press/v202/lyzhin23a.html}, abstract = {Nowadays, state-of-the-art learning-to-rank methods are based on gradient-boosted decision trees (GBDT). The most well-known algorithm is LambdaMART which was proposed more than a decade ago. Recently, several other GBDT-based ranking algorithms were proposed. In this paper, we thoroughly analyze these methods in a unified setup. In particular, we address the following questions. Is direct optimization of a smoothed ranking loss preferable over optimizing a convex surrogate? How to properly construct and smooth surrogate ranking losses? To address these questions, we compare LambdaMART with YetiRank and StochasticRank methods and their modifications. We also propose a simple improvement of the YetiRank approach that allows for optimizing specific ranking loss functions. As a result, we gain insights into learning-to-rank techniques and obtain a new state-of-the-art algorithm.} }
Endnote
%0 Conference Paper %T Which Tricks are Important for Learning to Rank? %A Ivan Lyzhin %A Aleksei Ustimenko %A Andrey Gulin %A Liudmila Prokhorenkova %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-lyzhin23a %I PMLR %P 23264--23278 %U https://proceedings.mlr.press/v202/lyzhin23a.html %V 202 %X Nowadays, state-of-the-art learning-to-rank methods are based on gradient-boosted decision trees (GBDT). The most well-known algorithm is LambdaMART which was proposed more than a decade ago. Recently, several other GBDT-based ranking algorithms were proposed. In this paper, we thoroughly analyze these methods in a unified setup. In particular, we address the following questions. Is direct optimization of a smoothed ranking loss preferable over optimizing a convex surrogate? How to properly construct and smooth surrogate ranking losses? To address these questions, we compare LambdaMART with YetiRank and StochasticRank methods and their modifications. We also propose a simple improvement of the YetiRank approach that allows for optimizing specific ranking loss functions. As a result, we gain insights into learning-to-rank techniques and obtain a new state-of-the-art algorithm.
APA
Lyzhin, I., Ustimenko, A., Gulin, A. & Prokhorenkova, L.. (2023). Which Tricks are Important for Learning to Rank?. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:23264-23278 Available from https://proceedings.mlr.press/v202/lyzhin23a.html.

Related Material