AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

The Weighted Generalised Covariance Measure

Cyrill Scheidegger, Julia Hörrmann, Peter Bühlmann

JMLR 2022 · Learning theory and algorithms › Kernels and regression theory · approach: Kernel two-sample testing · 9 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
use weighted GCM (WGCM.fix) instead of GCM for variable selectionconditional independence testing · Variable selection example (50 predictors, n=500)correct predictor set found out of 100: GCM never; WGCM.est 86 of 100; WGCM.fix 99 of 100Improved
use WGCM instead of GCMconditional independence test · Wave energy converters (Tasmania)significant predictors of 32: 0 -> 11 (WGCM.est), 0 -> 32 (WGCM.fix)Improved
use WGCM instead of GCMconditional independence test · Online news popularitysignificant variables: 8 -> 9 (WGCM.est), 8 -> 11 (WGCM.fix)Improved
use WGCM instead of GCMconditional independence test · Boston housingsignificant variables: 5 -> 3Hurt
use WGCM instead of GCM when dependence is nearly linear (high lambda)conditional independence test · Concrete example (n=200, 1000 runs, alpha=0.05)rejection rate (power)Hurt
use WGCM instead of GCM when dependence is nonlinear (lambda=0)conditional independence test · Concrete example (n=200, 1000 runs, alpha=0.05)rejection rate (power)Improved
use WGCM.fix (fixed weights) instead of WGCM.est (estimated weight)WGCM · Alternative hypothesis simulations (1D, 10D.Add, 10D.NonAdd)rejection rateImproved
use gam (additive regression) instead of xgboost for the nonadditive setting (ablation)GCM, WGCM.est, WGCM.fix · 10D.NonAdd null hypothesis (n=400)type I error rateHurt
increase sample size from 100 or 200 to 400 (ablation)GCM, WGCM.est, WGCM.fix · Null hypothesis simulations (1D, 10D.Add)type I error rate: n=100, 200 reject too often; n=400 not significantly above 0.05Improved

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