AI Paper Atlas

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

Leveraging Dark Knowledge for Intrinsic Multimodal Out-of-Distribution Detection

Nimeshika Udayangani Hewa Dehigahawattage, Sarah Erfani, Christopher Leckie

ECCV 2026 · Label-efficient and robust learning › OOD and anomaly detection · approach: Spurious-aware OOD detection · 0 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 self multimodal OOD distillationvanilla joint classifier · HMDB51 25/26 near-OODFPR95: 38.78 -> 34.99Improved
add self multimodal OOD distillationvanilla joint classifier · UCF101 50/51 near-OODFPR95: 10.10 -> 5.92Improved
add distillation frameworkBase · HMDB51 far-OODFPR95: 26.00 -> 21.88Improved
add distillation frameworkA2D · HMDB51 far-OODFPR95: 20.58 -> 17.54Improved
add distillation frameworkDPU · HMDB51 25/26 near-OODFPR95: 34.42 -> 33.25Improved
add distillation frameworkDPU · UCF101 50/51 near-OODFPR95: 7.57 -> 7.84Hurt
add distillation frameworkFM · HMDB51 far-OODAUROC: 94.12 -> 93.92No clear effect
use TCKD only (no NCKD) (ablation)self-distillation · HMDB51 25/26 near-OODFPR95: 38.78 -> 34.99Improved
use NCKD only (no TCKD) (ablation)self-distillation · HMDB51 far-OOD EPIC-KitchensFPR95: 22.12 -> 16.01Improved
weight modalities by likelihood instead of uniformly (ablation)distillation framework · HMDB51 far-OODFPR95: 22.96 -> 21.88Improved
use all modalities as teachers instead of best modality only (ablation)distillation framework · HMDB51 far-OODFPR95: 22.52 -> 21.88Improved
add logit distillation versus naive ensembling with feature fusion (ablation)joint classifier · HMDB51 far-OODFPR95: 26.97 -> 21.88Improved

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