Scalable Learning Of Non-decomposable Objectives

Abstract

Modern retrieval systems are often driven by an underlying machine learning model. The goal of such systems is to identify and possibly rank the few most relevant items for a given query or context. Thus, such systems are typically evaluated using a ranking-based performance metric such as the area under the precision-recall curve, the \(F_\beta\) score, precision at fixed recall, etc. Obviously, it is desirable to train such systems to optimize the metric of interest. In practice, due to the scalability limitations of existing approaches for optimizing such objectives, large-scale retrieval systems are instead trained to maximize classification accuracy, in the hope that performance as measured via the true objective will also be favorable. In this work we present a unified framework that, using straightforward building block bounds, allows for highly scalable optimization of a wide range of ranking-based objectives. We demonstrate the advantage of our approach on several real-life

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyeban2016scalable

Related papers

• Optimizing Compound Retrieval Systems (2025)0.00
• Optimization Of Rank Losses For Image Retrieval (2023)9.43
• Smooth-ap: Smoothing The Path Towards Large-scale Image Retrieval (2020)15.06
• Sodeep: A Sorting Deep Net To Learn Ranking Loss Surrogates (2019)12.33
• Scaling Laws For Dense Retrieval (2024)10.07
• Optimizing Retrieval Components For A Shared Backbone Via Component-wise Multi-stage Training (2026)0.00
• Pylate: Flexible Training And Retrieval For Late Interaction Models (2025)3.58
• Scalingnote: Scaling Up Retrievers With Large Language Models For Real-world Dense Retrieval (2024)0.00