AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

AP-OOD: Attention Pooling for Out-of- Distribution Detection

Claus Hofmann, Christian Huber, Bernhard Lehner, Daniel Klotz, Sepp Hochreiter, Werner Zellinger

ICLR 2026 · Label-efficient and robust learning › OOD and anomaly detection · approach: Spurious-aware OOD detection · 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
replace mean pooling with learned attention pooling over token embeddingsMahalanobis · XSUM summarization input OOD CNN/DMFPR95: 74.20 -> 12.88Improved
replace mean pooling with attention pooling (AP-OOD vs Mahalanobis)OOD detector · XSUM summarization input OOD (mean)AUROC: 88.44 -> 98.91Improved
replace mean pooling with attention poolingOOD detector · XSUM summarization output OOD (mean)AUROC: 88.49 -> 96.39Improved
use AP-OOD instead of KNNOOD detector · WMT15 En-Fr input OOD (mean)AUROC: 71.00 -> 74.81Improved
use AP-OOD instead of KNNOOD detector · WMT15 En-Fr output OOD (mean)AUROC: 74.39 -> 77.38Improved
use AP-OOD instead of MSPclassifier OOD detector · MIMII-DG audioFPR95: 36.43 -> 22.35Improved
add auxiliary OOD samples (semi-supervised AP-OOD)AP-OOD · XSUM summarizationAUROCImproved
use fully supervised AP-OOD instead of unsupervised AP-OODAP-OOD · XSUM summarization input OOD (mean)AUROC: 98.91 -> 99.94Improved
use AP-OOD instead of binary logits in supervised settingOOD detector · XSUM summarization input OOD (mean)AUROC: 99.74 -> 99.94Improved
scale AP-OOD to more heads and queries (M=1024, T=16) (ablation)AP-OOD · XSUM summarization (full XSUM training)AUROCImproved
use perplexity instead of embedding-based methods for output OODPerplexity · WMT15 En-Fr output OOD (mean)FPR95: 81.10 -> 77.08Improved

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