AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Seeing What’s Not There: Negation Understanding Needs More Than Training

Bhuvan Aggarwal, Amit More, Mudit Soni, S Divakar Bhat

ICLR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Language models for zero-shot reasoning

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
add embedding correction (negated-concept projection removal)CLIP · NegBench MCQ COCOMCQ accuracy: 24.7 -> 72.5Improved
add embedding correctionCLIP · NegBench MCQ VOC2007MCQ accuracy: 24.3 -> 78.6Improved
add embedding correctionCLIP · NegBench Retrieval COCOR-Neg@5: 57.3 -> 63.2Improved
add embedding correctionNegCLIP* · NegBench MCQ COCOMCQ accuracy: 56.2 -> 69.5Improved
add embedding correctionNegCLIP* · NegBench Retrieval MSRvttR-Neg@5: 51.5 -> 52.2Improved
add embedding correctionSigLIP2 · NegBench MCQ COCOMCQ accuracy: 27.2 -> 66.3Improved
add embedding correctionCLIP · VALSE Existenceaccuracy: 71.0 -> 79.5Improved
add embedding correctionCLIP · Flickr30K negation subsettext retrieval Acc R@1: 34.0 -> 41.3Improved
add embedding correctionCLIP · NegBench MCQ COCO (negated captions)confidence score: 0.16 -> 0.64Improved
use custom negation scope rules instead of off-the-shelf NegEx (ablation)Ours + CLIP · NegBench MCQ COCOMCQ accuracy: 61.3 -> 72.5Improved
use neutral anchor word instead of arbitrary noun (ablation)Ours + CLIP · NegBench MCQ COCOMCQ accuracy: 50.78 -> 72.51Improved
choose among neutral anchor words (neutral, balanced, unbiased, fair) (ablation)Ours + CLIP · NegBench MCQ COCOMCQ accuracy: 72.42 -> 72.54No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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