Reverse active learning based atrous DenseNet for pathological image classification.

The DRAL + ADN framework is a potential candidate for boosting the performance of deep learning models for partially mislabeled training datasets.