AI Paper Atlas

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

A Rigorous Information-Theoretic Definition of Redundancy and Relevancy in Feature Selection Based on (Partial) Information Decomposition

Patricia Wollstadt, Sebastian Schmitt, Michael Wibral

JMLR 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 21 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 CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman Itrue positive rate (%): 81.0 -> 100.0Improved
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman Ifalse negative rate (%): 18.4 -> 0.0Improved
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman IItrue positive rate (%): 56.8 -> 50.5Hurt
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman IIfalse positive rate (%): 0.8 -> 0.2Improved
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman IIItrue positive rate (%): 3.3 -> 3.5No clear effect
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Friedman (all three models)true positive rate (%)Mixed
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Runge et al. 2015 modeltrue positive rate (%): 54.7 -> 99.6Improved
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Runge et al. 2015 modelfalse positive rate (%): 7.5 -> 3.0Improved
use CMI instead of MI as inclusion criterion (ablation)sequential forward selection · Runge et al. 2015 modelfalse negative rate (%): 45.3 -> 0.4Improved
choose CMI forward selection over LARS/RF forward selection with fixed feature countfeature selection · Friedman II and IIIfalse positives from nuisance variablesImproved
select feature set by CMI criterion instead of using all features (ablation)KNN predictor (k=5) · Friedman IIMAEImproved
increase signal-to-noise by lowering noise sigmaPID/CMI estimates · Experiment II statistical modelsabsolute CMI, MI and PID valuesImproved

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