演講者：陳泓勳博士

　　　　國立交通大學大數據研究中心

日　期：2018年2月27日（星期二）15:30

地　點：國立高雄大學理學院408室

講　題：Applying transfer learning on medical image classification

摘　要：

In this talk, we introduce a practical case on medical image classification of using transfer learning. The main idea purpose of transfer learning is using learned-knowledge to solve new problems for better performance or less time consuming.

Inception is based on data from the large image database, ImageNet, to train models, which is a pre-trained CNN model provided by Google. Based on the inception v3 model, we extracted the characteristic from the spinal X-ray image of the spinal cord and added additional hidden layers for training to distinguish the spinal cord for compressive fractures.