AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Vector Linking via Cross-Model Local Isometric Consistency

Ziying Chen, Yang Cao, He Sun, Beining Yang, Tianjian Yang

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 0 citations (OpenAlex)

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
add bootstrapping of confident links as new anchors (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 1.9 -> 70.5Improved
add multi-view voting (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 29.8 -> 70.5Improved
add adaptive view scheduling (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 43.4 -> 70.5Improved
add FPS view sampling (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 20.9 -> 70.5Improved
add kernelized distance signature (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 51.0 -> 70.5Improved
add both FPS and kernel (ablation)GEH · SciDocs ablation (alpha 0.15, 15 seeds)F1 (%): 1.4 -> 70.5Improved
increase seed anchors from 15 to 30 (overlap .3) (ablation)GEH · SciFact (Mistral-OpenAI)F1 (%): 85.8 -> 85.3No clear effect
increase seed anchors from 15 to 30 (overlap .3) (ablation)Procrustes · SciFact (Mistral-OpenAI)F1 (%): 8.9 -> 40.1Improved
use GEH instead of CCA for linkingMistral-OpenAI · FEVER (alpha 0.3, 30 seeds)precision (%): 5.22 -> 93.8Improved
use kernelized encoding instead of raw distance (ablation)GEH · SciDocsprecision/recallImproved
use out-of-domain anchors instead of in-domainGEH · OOD seeds (30, overlap 30%)precision/recall (%): 77-87 precision, 80-97 recallNo clear effect

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