AI Paper Atlas

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

CLIPN for Zero-Shot OOD Detection: Teaching CLIP to Say No

Hualiang Wang, Yi Li, Huifeng Yao, Xiaomeng Li

ICCV 2023 · Label-efficient and robust learning › OOD and anomaly detection · approach: Spurious-aware OOD detection · 125 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 CLIPN-A (learnable no text encoder, agreeing-to-differ) versus MCMCLIP ViT-B-16 · ImageNet-1K zero-shot OODAUROC / FPR95: +2.34 AUROC, -11.64 FPR95Improved
add text semantic-opposite loss to image-text binary-opposite loss (ablation)CLIPN-A · ImageNet-1K OOD setsAUROC: iNaturalist 93.08 -> 95.27Improved
use learnable no prompts instead of handcrafted for competing-to-win (ablation)CLIPN-C · ImageNet-1K OOD setsAUROC: +6.83, +6.85, +0.66, +6.86Improved
use learnable no prompts instead of handcrafted for agreeing-to-differ (ablation)CLIPN-A · ImageNet-1K OOD setsAUROC: +2.52, +2.38, +2.63; Texture -1.22Mixed
use agreeing-to-differ instead of competing-to-win at inference (ablation)CLIPN · ImageNet-1K avg, ViT-B-16AUROC: 86.35 -> 93.10Improved
CLIPN-A versus best baselines on CIFAR-100 ID small-scaleCLIP ViT-B-32 · CIFAR-100 IDAUROC / FPR95: +5.23 AUROC avg; ODIN better on ImageNet_R and LSUN FPR95Mixed
use CLIPN-A instead of MSP thresholds on CIFAR-10 vs CIFAR-100CLIPN-A · CIFAR-10 ID, CIFAR-100 OODmisclassified OOD samples: 666/410/171 -> 45Improved
train only the no text encoder with frozen image and text encoders instead of full CLIP trainingCLIPN · training ViT-L-14 batch 16time per iteration / GPU memory: 0.58 -> 0.24 s; 18.91 -> 6.62 GImproved
add a second text encoder (no encoder) at inferenceCLIP ViT-B-32 · ImageNet-1K inferenceavg AUROC: 85.59 -> 90.53 (+4.94)Improved
use CLIPN-A on in-domain CIFAR splitsCLIP · CIFAR-10 / CIFAR-100 in-domainAUROC: +4.5 CIFAR10, +7.75 CIFAR100 over MSPImproved

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