AI Paper Atlas

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

Probing Classifiers are Unreliable for Concept Removal and Detection

Abhinav Kumar, Chenhao Tan, Amit Sharma

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 5 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
run INLP iterations on a clean classifierINLP null-space removal · Synthetic-Textmain-task accuracy: 100\% -> 50\%Hurt
higher predictive correlation kappaINLP null-space removal · Synthetic-Textspeed of main-task feature removalHurt
early stopping INLP when main performance dropsINLP null-space removal · Synthetic-Text kappa=0.8DeltaProb: 25\%Hurt
apply INLP to clean classifier on MultiNLIINLP null-space removal · MultiNLImain-task accuracyHurt
adversarial removal instead of ERMRoBERTa classifier · MultiNLIspuriousness scoreMixed
adversarial removal instead of ERMRoBERTa classifier · Twitter-PAN16, Twitter-AAEspuriousness scoreNo clear effect
adversarial removal instead of ERMRoBERTa classifier · Synthetic-Textspuriousness scoreHurt
feed clean classifier to adversarial removalAR · MultiNLIspuriousness scoreHurt

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