Adaptive Pseudo Label Selection for Individual Unlabeled Data by Positive and Unlabeled Learning
By: Takehiro Yamane , Itaru Tsuge , Susumu Saito and more
Potential Business Impact:
Helps doctors find sickness in X-rays better.
This paper proposes a novel pseudo-labeling method for medical image segmentation that can perform learning on ``individual images'' to select effective pseudo-labels. We introduce Positive and Unlabeled Learning (PU learning), which uses only positive and unlabeled data for binary classification problems, to obtain the appropriate metric for discriminating foreground and background regions on each unlabeled image. Our PU learning makes us easy to select pseudo-labels for various background regions. The experimental results show the effectiveness of our method.
Similar Papers
Cost-Sensitive Unbiased Risk Estimation for Multi-Class Positive-Unlabeled Learning
Machine Learning (CS)
Helps computers learn from good and unknown examples.
Learning from Double Positive and Unlabeled Data for Potential-Customer Identification
Machine Learning (CS)
Finds people likely to buy with less ads.
Heterogeneous Multisource Transfer Learning via Model Averaging for Positive-Unlabeled Data
Machine Learning (Stat)
Finds bad guys using less information.