AI Paper Atlas

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

Leveraging Evidence Priors for Robust Prompt Learning under Noisy Supervision in Vision-Language Models

Junnan Zou, Zhu Teng, Wei Zhang, Ming He, Jianping Fan

ICML 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Prompt learning for CLIP

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
co-supervise with pre-trained CLIP prior (equal weight)CoOp · OxfordPets symmetric 75%accuracy: 24.60 -> 66.50Improved
co-supervise with pre-trained CLIP prior (equal weight)CoOp · EuroSAT symmetric 75%accuracy: 26.70 -> 35.05Improved
replace naive prior with Evidence-PromptCoOp · OxfordPets symmetric 75%accuracy: 66.50 -> 88.83Improved
add marginal evidence (ablation)Evidence-Prompt · EuroSAT symmetric 75%accuracy: 37.11 -> 56.20Improved
add joint evidence (ablation)Evidence-Prompt · EuroSAT symmetric 75%accuracy: 50.23 -> 56.20Improved
add E_L^nl loss term (ablation)Evidence-Prompt · EuroSAT symmetric 75%accuracy: 38.17 -> 56.20Improved
add E_L^sem loss term (ablation)Evidence-Prompt · EuroSAT symmetric 75%accuracy: 53.54 -> 56.20Improved
add E_L^sem loss term (ablation)Evidence-Prompt · EuroSAT asymmetric 75%accuracy: 11.85 -> 34.84Improved
add joint evidence (ablation)Evidence-Prompt · EuroSAT asymmetric 75%accuracy: 10.60 -> 34.84Improved
add marginal evidence (ablation)Evidence-Prompt · EuroSAT asymmetric 75%accuracy: 35.23 -> 34.84No clear effect
plug evidence prior into JoAPRJoAPR · OxfordPets symmetric 75%accuracy: 74.40 -> 85.99Improved
plug evidence prior into NLPromptNLPrompt · OxfordPets symmetric 75%accuracy: 70.77 -> 86.43Improved

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