AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Leveraging Data to Say No: Memory Augmented Plug-and-Play Selective Prediction

Aditya Sarkar, Yi Li, Jiacheng Cheng, Shlok Kumar Mishra, Nuno Vasconcelos

ICLR 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 1 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
add memory retrieval (MA-PaPSP) to PaPSPPaPSP · Pets classificationAURC: 0.169 -> 0.076Improved
add memory retrieval (MA-PaPSP) to PaPSPPaPSP · MS-COCO captioningAURC Cider-N: 0.136 -> 0.126Improved
add memory retrieval (MA-PaPSP) to PaPSPPaPSP · Flickr-30K captioningAURC Cider-N: 0.229 -> 0.229No clear effect
add memory retrieval (MA-PaPSP) to PaPSPPaPSP · VL-checklist ITMAURC: 0.204 -> 0.207Hurt
add memory retrieval (MA-PaPSP) to PaPSPPaPSP · UCF-101 classificationAURC: 0.113 -> 0.078Improved
fine-tune SP-VLM on in-domain data for PaPSPPaPSP · Pets classificationAURC: 0.169 -> 0.074Improved
fine-tune SP-VLM on in-domain data for MA-PaPSPMA-PaPSP · MS-COCO captioningAURC Meteor: 0.288 -> 0.234Improved
fine-tune SP-VLM on in-domain data for MA-PaPSPMA-PaPSP · Foil ITMAURC: 0.167 -> 0.162Hurt
add proxy embedding and contrastive score to PaPSP (ablation)PaPSP · captioning, ITM, classificationAURCImproved
use soft nearest neighbor proxy (K>1) instead of top-1 retrieval (ablation)MA-PaPSP · captioning, ITM, classificationAURCImproved
increase neighborhood size K beyond about 15 (ablation)MA-PaPSP · captioning, ITM, classificationAURCHurt
use image-to-text retrieval (i2tr) instead of other retrieval variants (ablation)MA-PaPSP · captioning, ITM, classificationAURCImproved

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