AI Paper Atlas

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

Beyond Top Activations: Efficient and Reliable Crowdsourced Evaluation of Automated Interpretability

Tuomas Oikarinen, Ge Yan, Akshay Kulkarni, Tsui-Wei Weng

CVPR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Post-hoc explanation analysis · 0 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 model-guided importance sampling instead of uniform sampling (ablation)correlation estimation · ImageNet simulationsamples to match accuracy: ~15x fewer samples (abstract: ~13x)Improved
use model-guided importance sampling instead of uniform sampling (ablation)correlation estimation · ImageNet simulationRCE at same budget: ~65% lower errorImproved
add BRAgg (SigLIP prior) rating aggregation instead of majority vote (ablation)correlation estimation · simulation, 23% noiseraters per input: 2-10x fewer ratersImproved
use average aggregation instead of Bayes aggregation (ablation)correlation estimation · simulation, 23% noiseRCE: fails to reach low error even at 100 ratersHurt
add BRAgg to MG-IS (ablation)correlation estimation · simulation, 550 evalsevaluations to reach 27.5% RCE: 1760 -> 550Improved
add MG-IS to BRAgg (ablation)correlation estimation · simulationevaluations to reach 27.5% RCE: 14607 -> 550Improved
combine MG-IS and BRAgg vs uniform + majority vote (ablation)correlation estimation · simulationevaluations to reach 27.5% RCE: 22560 -> 550Improved
add BRAgg to MG-IS (ablation)correlation estimation · MTurk validationevaluations to reach 19.8% RCE: 2000 -> 550Improved
use human study (MG-IS+BRAgg) instead of SigLIP-only evaluationcorrelation estimation · all neuronsRCE: 32.10% -> 27.90%Improved
use Linear Explanations instead of LLM-based methods (MAIA, DnD)neuron explanation · ViT-B-16 layer11 MLP, SigLIP simcorrelation: 0.1497 -> 0.2968Improved
use LLM-based explanations (DnD, MAIA) instead of simpler CLIP-Dissectneuron explanation · ResNet-50 layer4, SigLIP simcorrelation: 0.1904 -> 0.1534 (DnD)Hurt
use LLM agent MAIA instead of CLIP-Dissectneuron explanation · ViT-B-16 layer11 MLP, SigLIP simcorrelation: 0.1849 -> 0.1497Hurt

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