AI Paper Atlas

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

Confidence Sets with Expected Sizes for Multiclass Classification

Christophe Denis, Mohamed Hebiri

JMLR 2017 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 14 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
replace single-label prediction by random forest beta-set with beta=2random forest · Plant (K=100)risk: 0.40 -> 0.18Improved
aggregate library by superlearning instead of best single beta-setconfidence sets · Plant (K=100), beta=2risk R: 0.20 -> 0.17Improved
use CV aggregation instead of softmax regression beta-setconfidence sets · Plant (K=100), beta=2risk R: 0.77 -> 0.17Improved
increase beta from 2 to 10CV confidence set · Plant (K=100)risk R: 0.17 -> 0.02Improved
increase beta from 2 to 10random forest beta-set · Plant (K=100)risk R: 0.18 -> 0.02Improved
use CV aggregation instead of best library componentsuperlearning CV · Forest (K=4), beta=2risk R: 0.02 -> 0.02No clear effect
use CV aggregation instead of softmax regression beta-setconfidence sets · Forest (K=4), beta=2risk R: 0.06 -> 0.02Improved
calibrate threshold with G estimated from unlabeled databeta-set · Forest and Plantinformation IImproved

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