LINe (ResNet-50)

Reported on 6 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision6 results

Out-of-Distribution DetectiononImageNet-1k vs iNaturalist
AUROC· 2023-03-24
97.56
best: 99.62 (CMA(ViT-B/16, NegLabel))
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995
Out-of-Distribution DetectiononImageNet-1k vs iNaturalist
FPR95· 2023-03-24
12.26
best: 68.32 (GEN)
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995
Out-of-Distribution DetectiononImageNet-1k vs Textures
AUROC· 2023-03-24
94.44
best: 98.92 (ViM (BiT))
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995
Out-of-Distribution DetectiononImageNet-1k vs Textures
FPR95· 2023-03-24
22.54
best: 79.8 (RP+GradNorm)
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995
Out-of-Distribution DetectiononImageNet-1k vs Curated OODs (avg.)
AUROC· 2023-03-24
95.03
best: 95.71 (SCALE (ResNet50))
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995
Out-of-Distribution DetectiononImageNet-1k vs Curated OODs (avg.)
FPR95· 2023-03-24
20.7
best: 70.12 (RP+GradNorm)
LINe: Out-of-Distribution Detection by Leveraging Important Neurons arXiv:2303.13995