Pretrained deep models outperform GBDTs in Learning-To-Rank under label scarcity

Charlie Hou, Kiran Koshy Thekumparampil, Michael Shavlovsky, Giulia Fanti, Yesh Dattatreya, sujay sanghavi

Abstract: On tabular data, a significant body of literature has shown that current deep learning (DL) models perform at best similarly to Gradient Boosted Decision Trees (GBDTs), while significantly underperforming them on outlier data. However, these works often study problem settings which may not fully capture the complexities of real-world scenarios. We identify a natural tabular data setting where DL models can outperform GBDTs: tabular Learning-to-Rank (LTR) under label scarcity. Tabular LTR applications, including search and recommendation, often have an abundance of unlabeled data, and scarce labeled data. We show that DL rankers can utilize unsupervised pretraining to exploit this unlabeled data. In extensive experiments over both public and proprietary datasets, we show that pretrained DL rankers consistently outperform GBDT rankers on ranking metrics, sometimes by as much as 38%, both overall and on outliers.

Submission Length: Regular submission (no more than 12 pages of main content)

Changes Since Last Submission: Addressed the reviewer comments by revising the text and adding new experiments.

Code: https://github.com/houcharlie/ltr-pretrain/

Assigned Action Editor: ~Hector_Palacios1

Submission Number: 2923