AI Paper Atlas

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

A flexible model-free prediction-based framework for feature ranking

Jingyi Jessica Li, Yiling Elaine Chen, Xin Tong

JMLR 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Feature attribution methods · 4 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 s-NPC with small alpha .01 instead of s-CCfeature ranking · Toy examplefrequency feature 2 ranked top: 22.0% -> 98.4%Improved
raise NP level alpha from .01 to .20s-NPC · Toy examplefrequency feature 1 ranked top: 1.6% -> 99.0%Mixed
use s-NPC or s-CC instead of Pearson correlationmarginal ranking · S6 two-feature modelfrequency recovering true rank: 0% -> 100%Improved
use s-CC instead of two-sample Wilcoxonmarginal ranking · S6 two-feature modelfrequency recovering true rank: 0% -> 100%Improved
use s-NPC (alpha=.10) instead of distance correlationmarginal ranking · S6 two-feature modelfrequency recovering true rank: 0.5% -> 99.3%Improved
increase sample size from 400 to 1000s-CC and s-NPC · S1 Gaussianagreement with true ranksImproved
decrease alpha (e.g. to 0.05) (ablation)s-NPC · S1 Gaussian N=400rank variabilityHurt
use s-NPC instead of s-CC or SVM variants under sampling bias (class 1 proportion .5 -> .1)feature ranking · S3 Gaussian, sampling biasseparating 10 informative featuresImproved
use s-NPC instead of its SVM variants-NPC · S5 d=10000, N=200ranking accuracy of top 10 featuresImproved
use alpha=.30 instead of .10 or .20s-NPC · S5 d=10000, N=200average rank of 10 informative featuresImproved
use larger alpha .2 or .3 instead of .05 or .1s-NPC · S4 d=500, N=400ranking features 8-10Improved
use mean decrease in accuracy instead of mean decrease in Ginirandom forest importance · S5 d=10000, N=200rank of 10th featureImproved

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