Similarity-based Active Learning for Image Classification under Class Imbalance

摘要

Many image classification tasks (e.g., medical image classification) have a severe class imbalance problem. Convolutional neural network (CNN) is currently a state-of-the-art method for image classification. CNN relies on a large training dataset to achieve high classification performance. However, manual labeling is costly and may not even be feasible for the medical domain. In this paper, we propose a novel similarity-based active deep learning framework (SAL) that deals with class imbalance. SAL actively learns a similarity model to recommend unlabeled rare class samples for experts' manual labeling. Based on similarity ranking, SAL recommends with high confidence unlabeled common class samples for automatic pseudo-labeling without requiring labeling effort by experts. Our experiments show that SAL consistently outperforms two other recent active deep learning methods on two challenging datasets. In addition, SAL obtains nearly the upper bound classification performance (using all the images in the training dataset) for an Endoscopy and the Caltech-256 datasets while the domain experts labeled only 5.6% and 7.5% of all images, respectively. To the best of our knowledge, SAL is the first active deep learning framework that deals with a significant class imbalance and significantly reduces manual labeling efforts by experts while achieving near optimal classification performance.