Mid-Level Feature Extractor for Transfer Learning to Small-Scale Dataset of Medical Images

초록

In fine-tuning-based transfer learning, the size of the dataset may affect the learning accuracy. When a dataset scale is small, fine-tuning-based transfer learning methods use high computing costs, similar to a large-scale dataset. we propose a mid-level feature extractor that only retrains the mid-level convolutional layers, resulting in increased efficiency and reduced computing costs. This mid-level feature extractor is likely to provide an effective alternative in training a small-scale medical image dataset. The performance of the mid-level feature extractor is compared with performance of low- and high-level feature extractors, as well as the fine-tuning method. The mid-level feature extractor takes shorter time to converge than other methods, and it shows good accuracy, obtaining an area under the ROC curve (AUC) of 0.87 in untrained test dataset that is very different from training dataset. © 2020, Springer Nature Singapore Pte Ltd.

키워드