Article

Activation Fluctuation-Driven Scaling for Reliable Medical Out-of-Distribution Detection

2026-05-31

Abstract excerpt

<title>Abstract</title> <p>Reliable out-of-distribution (OOD) detection is essential for safely deploying deep learning models in open-world medical imaging scenarios, where unexpected inputs may lead to high-risk diagnostic errors. Scale, a representative post-hoc method, enlarges the distributional discrepancy between in-distribution (ID) and OOD samples by globally scaling neural activations. However, it does...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: cdaef0c8-aef7-576f-8be7-628f193e2d65
DOI: 10.21203/rs.3.rs-9608616/v1

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

Activation Fluctuation-Driven Scaling for Reliable Medical Out-of-Distribution DetectionDOI 10.21203/rs.3.rs-9608616/v1Read

Beyond Transfer Learning: A Generative Self-Supervised Framework for fMRI-Based Diagnosis on Small and Imbalanced DatasetsSemantic proximity Read

Scalable Drift Monitoring in Medical Imaging AISemantic proximity Read

A Systematic Review of Cross-Population Shifts in Medical Imaging Analysis with Deep LearningSemantic proximity Read

Uncertainty in Deep Learning for EEG under Dataset ShiftsSemantic proximity Read

Brain Cancer, MRI, Deep Learning, Generalization, Overfitting, Image ClassificationSemantic proximity Read

Domain Shift-Robust Tumor Segmentation Across HospitalsSemantic proximity Read

Dynamic Incremental Learning in Medical ImagingSemantic proximity Read

CCS: A Continuous Spatial-Semantic Concordance Score for Robust Evaluation of Object Detection ModelsSemantic proximity Read

A Survey of Generalization and Adaptation in Medical Imaging Foundation ModelsSemantic proximity Read

WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical ImagingSemantic proximity Read

CLSENet: A Continuous Learning Semantic Enhancement Network for Medical Image PerceptionSemantic proximity Read

CUPID-Med: Conformal Uncertainty-aware Prototype Discovery for Clinically Interpretable Deep Medical ClusteringSemantic proximity Read