AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Adversarial Reciprocal Points Learning for Open Set Recognition

Guangyao Chen, Peixi Peng, Xiangqian Wang, Yonghong Tian

TPAMI 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 394 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 confusing samples with adversarial BN and fine-tuning to ARPLARPL · TinyImageNet (open set)OSCR: 62.3 -> 65.9Improved
add confusing samples to ARPLARPL · TinyImageNet (open set)AUROC: 76.2 -> 78.2Improved
use ARPL instead of RPLRPL · CIFAR+10 (open set)AUROC: 84.2 -> 96.5Improved
use ARPL instead of RPLRPL · CIFAR10 (open set)AUROC: 82.7 -> 90.1Improved
use ARPL instead of softmax cross-entropysoftmax · TinyImageNet (open set)OSCR: 60.8 -> 62.3Improved
use GCPL prototypes instead of softmaxsoftmax · TinyImageNet (open set)OSCR: 60.8 -> 59.3Hurt
add adversarial batch norm (ABN) to ARPL+CS (ablation)ARPL+CS · OOD CIFAR10 vs SVHNAUROC: 90.9 -> 96.6Improved
add fine-tuning (FT) to ARPL+CS (ablation)ARPL+CS · OOD CIFAR10 vs SVHNAUROC: 95.6 -> 96.6Improved
add fine-tuning (FT) to ARPL+CS (ablation)ARPL+CS · OOD CIFAR10 vs CIFAR100AUROC: 89.7 -> 90.3Improved
add adversarial batch norm (ABN) to ARPL+CS (ablation)ARPL+CS · OOD CIFAR10 vs CIFAR100AUROC: 88.8 -> 90.3Improved
add FT to ARPL+CS (ablation)ARPL+CS · CIFAR100 (closed set)accuracy: 71.5 -> 72.8Improved
add ABN to ARPL+CS (ablation)ARPL+CS · CIFAR100 (closed set)accuracy: 71.8 -> 72.8Improved

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