AI Paper Atlas

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

Aggregated Hold-Out

Guillaume Maillard, Sylvain Arlot, Matthieu Lerasle

JMLR 2021 · Learning theory and algorithms › Prediction and decision losses · approach: Consistency of surrogate losses · 7 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
increase the number of splits V (ablation)Agghoo · epsilon-regressionexcess risk: most improvement between V = 1 and V = 5Improved
increase V beyond about 5-10 (ablation)Agghoo · epsilon-regression (tau >= 0.5)excess risk: V much larger seems uselessNo clear effect
increase training fraction tau from 0.1 to 0.5 (ablation)Agghoo · epsilon-regressionexcess risk: 0.1 to 0.5Improved
increase tau above 0.8 (ablation)Agghoo · epsilon-regressionexcess risk: tau > 0.8Hurt
aggregate several hold-out splits instead of single hold-outAgghoo · epsilon-regressionexcess riskImproved
replace CV by Agghoo with V >= 5 and tau >= 0.5Agghoo · epsilon-regressionexcess risk: V >= 5Improved
replace CV by Agghoo with V = 1Agghoo · epsilon-regressionexcess risk: V = 1Hurt
use aggregation instead of selectionAgghoo · epsilon-regression (SVR)excess risk vs oracleImproved
aggregate by majority vote instead of selectionMajhoo · k-NN classificationexcess risk vs oracle: at least 4 times larger than oracle 0.0034No clear effect
replace CV by Majhoo with V >= 10Majhoo · k-NN classificationexcess risk: V >= 10Improved
replace CV by Majhoo with V = 2Majhoo · k-NN classificationexcess risk: V = 2Hurt
increase tau above 0.6 (ablation)Majhoo · k-NN classificationexcess risk: tau > 0.6Hurt

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