AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Optimization of Rank Losses for Image Retrieval

Elias Ramzi, Nicolas Audebert, Clément Rambour, André de Melo Araújo, Xavier Bitot, Nicolas Thome

TPAMI 2025 · Learning theory and algorithms › Prediction and decision losses · approach: Ranking-based loss optimization

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace sigmoid rank with SupRank (ablation)Smooth-AP · SOPmAP@R: 54.3 -> 54.8Improved
replace sigmoid rank with SupRank (ablation)Smooth-AP · iNaturalistmAP@R: 26.5 -> 27.5Improved
add proxy-based decomposability loss L_DG* (ablation)Sup-AP · iNaturalistR@1: 68.9 -> 71.8Improved
add proxy-based decomposability loss L_DG* (ablation)Sup-R@k · iNaturalistR@1: 68.2 -> 71.9Improved
use proxy-based L_DG* instead of pair-based L_DG (ablation)ROADMAP · iNaturalistR@1: 69.1 -> 71.8Improved
use proxy-based L_DG* instead of pair-based L_DG (ablation)ROADMAP · SOPR@1: 81.7 -> 81.9No clear effect
reduce batch size from 384 to 32 when adding L_DG* (ablation)Sup-AP · iNaturalistrelative mAP@R increase: 3.5 -> 22.8Improved
increase lambda from 0.1 to 0.3 (ablation)ROADMAP · iNaturalistR@1: 71.8 -> 72.1Improved
increase slope rho in SupRank from 0 to above 1 (ablation)Sup-AP · iNaturalistmAP@RImproved
use ROADMAP instead of cross-batch memory XBMXBM · iNaturalistR@1: 59.3 -> 71.8Improved
optimize H-AP (HAPPIER) instead of CSLCSL · SOPH-AP: 52.8 -> 59.4Improved
optimize H-AP with hierarchy instead of fine-grained Smooth-APSmooth-AP · SOPH-AP: 42.9 -> 59.4Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy