SmoothI: Smooth Rank Indicators for Differentiable IR Metrics - Université Grenoble Alpes Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2021

SmoothI: Smooth Rank Indicators for Differentiable IR Metrics

Résumé

Information retrieval (IR) systems traditionally aim to maximize metrics built on rankings, such as precision or NDCG. However, the non-differentiability of the ranking operation prevents direct optimization of such metrics in state-of-the-art neural IR models, which rely entirely on the ability to compute meaningful gradients. To address this shortcoming, we propose SmoothI, a smooth approximation of rank indicators that serves as a basic building block to devise differentiable approximations of IR metrics. We further provide theoretical guarantees on SmoothI and derived approximations, showing in particular that the approximation errors decrease exponentially with an inverse temperature-like hyperparameter that controls the quality of the approximations. Extensive experiments conducted on four standard learning-to-rank datasets validate the efficacy of the listwise losses based on SmoothI, in comparison to previously proposed ones. Additional experiments with a vanilla BERT ranking model on a text-based IR task also confirm the benefits of our listwise approach.

Dates et versions

hal-03369664 , version 1 (07-10-2021)

Identifiants

Citer

Thibaut Thonet, Yagmur Gizem Cinar, Éric Gaussier, Minghan Li, Jean-Michel Renders. SmoothI: Smooth Rank Indicators for Differentiable IR Metrics. 2021. ⟨hal-03369664⟩
44 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More