.png){kind=icon}
返回
OSA-PD: Open-Set Active Learning Exploiting Predictive Discrepancy
.png){kind=icon}
PRE
AI
DOI：10.1109/LSP.2025.3537398.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Active learning with closed-set annotation has achieved significant success. However, real-world data often comprises numerous unknown classes irrelevant to the task which often confuses the query strategies to select unknown class data for annotation. To address such challenge in the open-set annotation (OSA), we propose a novel open-set active learning framework based on predictive discrepancy, named OSA-PD, with consideration that valuable data for model improvement is often with high predictive discrepancy. Specifically, two algorithms based on different discrepancy measurements are presented under OSA-PD framework, i.e., OSA-PRD with predictive results discrepancy, and OSA-PDD with decoupled predictive distribution discrepancy. Experimental results on CIFAR100 and TinyImageNet demonstrate the proposed OSA-PD can effectively select known class data and achieve higher classification accuracy with the same amount of annotated sample in comparison with existing active learning algorithms.
Keyword:
Detectors
Annotations
Active learning
Data models
Accuracy
Probability distribution
Prediction algorithms
Uncertainty
Training
Signal processing algorithms
open-set annotation
predictive discrepancy
image classification
期刊
IF:
9.6
论文数:
1.1W
被引数:
1.7W
机构
引用论文
Associative nitrogen fixation (ANF) in switchgrass (Panicum virgatum) across a nitrogen input gradient.png)
PLOS ONE
OAAI
PLOS ONEIF0
OAAI分享收藏Notes on New or Interesting Asplenium Species from Western Asia, Including Comments on Ching & Wu (1985), and Fraser-Jenkins (1992) Reliquiae Reichsteinianae 1关于西亚新或有趣铁角蕨属物种的笔记，包括对Ching & Wu (1985)以及Fraser-Jenkins (1992) Reliquiae Reichsteinianae 1的评论
PREAI
分享收藏
查看更多内容