AI Paper Atlas

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

PRISM: Prototype-based Reasoning with Inter-modal Semantic Mining for Interpretable Image Recognition

Anni Yu, Yu-Bin Yang

CVPR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Prototype and concept dictionary models

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 text-guided alignment loss Lalign (ablation)PRISM (Lce + Lorth) · CUB-200-2011, ViT-B/16consistency (%): 13.5 -> 96.3Improved
add Lalign (ablation)PRISM (Lce + Lorth) · CUB-200-2011, ViT-B/16stability (%): 51.2 -> 77.5Improved
add Lalign (ablation)PRISM (Lce + Lorth) · CUB-200-2011, ViT-B/16top-1 accuracy (%): 87.4 -> 87.5No clear effect
add compactness loss Lcomp on top of Lalign (ablation)PRISM · CUB-200-2011, ViT-B/16consistency (%): 96.3 -> 75.8Hurt
add compactness loss Lcomp on top of Lalign (ablation)PRISM · CUB-200-2011, ViT-B/16top-1 accuracy (%): 87.5 -> 87.8Improved
add Lcomp alone without Lalign (ablation)PRISM (Lce + Lorth) · CUB-200-2011, ViT-B/16top-1 accuracy (%): 87.4 -> 86.8Hurt
add Lcomp alone without Lalign (ablation)PRISM (Lce + Lorth) · CUB-200-2011, ViT-B/32top-1 accuracy (%): 82.2 -> 80.2Hurt
increase prototypes per class from 10 to 20 (ablation)PRISM · CUB-200-2011, ViT-B/16top-1 accuracy (%): 87.8 -> 88.5Improved
increase prototype dimension from 160 to 200 (ablation)PRISM · CUB-200-2011, ViT-B/16top-1 accuracy (%): 87.8 -> 88.7Improved
add CLIP-based semantic guidance to ProtoPNet prototype learningProtoPNet · CUB-200-2011 and Stanford Dogstop-1 accuracy: +2.3% to +4.6% on CUB, +1.9% to +3.4% on DogsImproved
use PRISM instead of concept bottleneck models using CLIP and GPT-3 conceptsLabel-Free CBM and FVLC · CUB-200-2011, ViT-B/16accuracy (%): 76.4 -> 87.8 and 77.9 -> 87.8Improved
add PRISM prototype alignmentProtoPNet · CUB-200-2011, ViT-B/16prototype consistency (%): 19.85 -> 75.75Improved

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