Skip to main content
Fig. 6 | BMC Medical Imaging

Fig. 6

From: A predicted-loss based active learning approach for robust cancer pathology image analysis in the workplace

Fig. 6

Example of the distribution of predicted loss values (a) Distribution types when predicting the loss with a trained LPM on the entire unlabeled dataset. Most of the noisy data aligns with the high-loss location similar to informative data. b Distribution obtained by predicting the loss with the same trained LPM as in (a) and classifying it into predicted classes. Within these predicted classes, it can be observed that noisy data are distributed in the higher loss location compared to informative data

Back to article page

BMC Medical Imaging

ISSN: 1471-2342

Contact us