AI Paper Atlas

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

Data-IQ: Characterizing subgroups with heterogeneous outcomes in tabular data

Nabeel Seedat, Jonathan Crabbé, Ioana Bica, Mihaela van der Schaar

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 6 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
Apply Group-DRO using Data-IQ groups instead of the baselineProstate model · Prostateoverall performance: 0.837 -> 0.840Improved
Apply Group-DRO using Data-IQ groups instead of the baselineCovid model · Covidoverall performance: 0.729 -> 0.732Improved
Apply Group-DRO using Data-IQ groups instead of the baselineSupport model · Supportoverall performance: 0.734 -> 0.734No clear effect
Apply Group-DRO using Data-IQ groups instead of the baselineFetal model · Fetaloverall performance: 0.768 -> 0.833Improved
Use Group-DRO (George) instead of Group-DRO with Data-IQ groupstabular medical models · Prostate, Covid, Support, and Fetaloverall performance: 0.840 vs 0.565; 0.732 vs 0.687; 0.734 vs 0.660; 0.833 vs 0.829Hurt
Use JTT instead of Group-DRO with Data-IQ groupstabular medical models · Prostate, Covid, Support, and Fetaloverall performance: 0.840 vs 0.723; 0.732 vs 0.455; 0.734 vs 0.611; 0.833 vs 0.829Hurt
Remove ambiguous training examplesprostate cancer model trained in one country · Cross-country generalization, SEER and CUTRACTtest-time generalization performanceImproved
Use aleatoric uncertainty rather than epistemic uncertainty to characterize examplesData-IQ vs Data Maps · Four medical datasetsSpearman rank correlation across model combinationsImproved

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