AI Paper Atlas

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

Calibratable Disambiguation Loss for Multi-Instance Partial-Label Learning

Wei Tang, Yi Yang, Weijia Zhang, Min-Ling Zhang

TPAMI 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 0 citations (OpenAlex)

Read the paper

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 CDL to MIPL methodsDEMIPL, ELIMIPL, MIPLMA · benchmark datasets, r=1..3accuracy: outperform in 67 of 72 casesImproved
add CDLMIPL methods · benchmark datasetsaccuracy: 3.27%, 4.57%, 8.93% mean gain for r=1,2,3Improved
add CDL-CC to DEMIPLDEMIPL · MNIST-MIPL r=3accuracy: 70.93 -> 85.27Improved
add CDL-CN to MIPLMAMIPLMA · MNIST-MIPL r=3accuracy: 74.93 -> 98.00Improved
add CDL-CN to DEMIPLDEMIPL · SIVAL-MIPL r=3accuracy: 50.31 -> 68.89Improved
add CDL-CN to DEMIPLDEMIPL · C-KMeansaccuracy: 52.11 -> 64.78Improved
add CDL-CC to ELIMIPLELIMIPL · FMNIST-MIPL r=1accuracy: 90.27 -> 90.27No clear effect
add CDL-CC to DEMIPLDEMIPL · MNIST-MIPL r=1ECE (%): 59.40 -> 1.08Improved
add CDL-CN to MIPLMAMIPLMA · FMNIST-MIPL r=1ECE (%): 52.38 -> 4.18Improved
add CDL-CN to MIPLMAMIPLMA · SIVAL-MIPL r=1ECE (%): 61.67 -> 15.91Improved
compare FastMIPL to CDL variantsFastMIPL · benchmark datasetsaccuracy: FastMIPL better in 4 of 72 casesHurt
add CDL-CC to DEMIPLDEMIPL · C-SIFTaccuracy: 53.17 -> 53.59No 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