AI Paper Atlas

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

Low-Dimensional Gradient Helps Out-of-Distribution Detection

Yingwen Wu, Tao Li, Xinwen Cheng, Jie Yang, Xiaolin Huang

TPAMI 2024 · Label-efficient and robust learning › OOD and anomaly detection · approach: Spurious-aware OOD detection · 13 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 feature input with low-dimensional gradientMSP · ImageNet, ResNet50FPR95: 64.76 -> 32.38Improved
replace feature input with low-dimensional gradientEnergy · ImageNet, ResNet50FPR95: 57.47 -> 27.90Improved
replace feature input with low-dimensional gradientReAct · ImageNet, ResNet50FPR95: 30.70 -> 23.03Improved
replace feature input with low-dimensional gradientKNN · ImageNet, ResNet50FPR95: 56.30 -> 43.43Improved
replace feature input with low-dimensional gradientMahalanobis · ImageNet, ResNet50FPR95: 81.51 -> 87.30Hurt
replace feature input with low-dimensional gradientReAct · ImageNet, ResNet50, SUN OODFPR95: 24.07 -> 45.34Hurt
replace feature input with low-dimensional gradientKNN · ImageNet, MobileNet-v2FPR95: 86.50 -> 65.32Improved
replace feature input with low-dimensional gradientEnergy · ImageNet, DenseNet121FPR95: 50.41 -> 50.56No clear effect
add forward-backward score ensemble (ablation)gradient-based ReAct · ImageNet, ResNet50FPR95: 23.03 -> 19.55Improved
add forward-backward score ensemble (ablation)gradient-based KNN · CIFAR10, ResNet18FPR95: 21.55 -> 20.24Improved
use 200-dim PCA subspace instead of 10-dim PCA (ablation)gradient + KNN · CIFAR10, ResNet18FPR95: 36.12 -> 21.55Improved
use 10-dim PCA subspace instead of 10-dim average-gradient subspace (ablation)gradient + KNN · CIFAR10, ResNet18FPR95: 39.14 -> 36.12Improved

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