.png){kind=icon}
返回
Uncertainty-Based Selective Clustering for Active Learning
.png){kind=icon}
OA
AI
DOI：10.1109/ACCESS.2022.3216065.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Labeling large amount of data is one of important issues in deep learning due to high labeling cost. One method to address this issue is the use of active learning. Active learning selects from a large unlabeled data pool, a set of data that is more informative to training a model for the task at hand. Many active learning approaches use uncertainty-based methods or diversity-based methods. Both have had good results. However, using uncertainty-based methods, there is a risk that sampled data may be redundant, and the use of redundant data can adversely affect the training of the model. Diversity-based methods risk losing some data important for training the model. In this paper, we propose the uncertainty-based Selective Clustering for Active Learning (SCAL), a method of selectively clustering for data with high uncertainty to sample data from each cluster to reduce redundancy. SCAL is expected to extend the area of the decision boundary represented by the sampled data. SCAL achieves a cutting-edge performance for classification tasks on balanced and unbalanced image datasets as well as semantic segmentation tasks.
Keyword:
Labeling
Data models
Uncertainty
Training
Learning systems
Deep learning
Clustering methods
Active learning
data sampling
uncertainty-based method
diversity-based method
pool-based scenario
期刊
IF:
3.6
论文数:
9.8W
被引数:
29.4W
机构
引用论文
Validation, comparison, and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images: The LUNA16 challenge计算机断层扫描图像中自动检测肺结节的算法的验证，比较和组合: LUNA16挑战.png)
MEDICAL IMAGE ANALYSIS
OAAI
MEDICAL IMAGE ANALYSISIF11.8
OAAI分享收藏Mechanisms of Disease: local renin–angiotensin–aldosterone systems and the pathogenesis and treatment of cardiovascular disease疾病机制: 局部肾素-血管紧张素-醛固酮系统与心血管疾病的发病机制和治疗
PREAI
分享收藏Foundations for Restraining Bolts: Reinforcement Learning with LTLf/LDLf Restraining SpecificationsOAAI
OAAI分享收藏
查看更多内容