On automated source selection for transfer learning in convolutional neural networks

作者：

Highlights：

• This study demonstrates that automatically ranking pre-trained source CNNs for a given target task, is possible.

• This study presents an information theoretic framework to rank source CNNs in an efficient, reliable, and zero-shot manner.

• This study presents a thorough experimental evaluation of the proposed theory using the Places-MIT database, CalTech-256 database, MNIST database and a real-world MRI database.

摘要

•This study demonstrates that automatically ranking pre-trained source CNNs for a given target task, is possible.•This study presents an information theoretic framework to rank source CNNs in an efficient, reliable, and zero-shot manner.•This study presents a thorough experimental evaluation of the proposed theory using the Places-MIT database, CalTech-256 database, MNIST database and a real-world MRI database.

论文关键词：Transfer learning,CNN selection,Deep learning

论文评审过程：Received 13 January 2017, Revised 23 April 2017, Accepted 17 July 2017, Available online 26 July 2017, Version of Record 18 September 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.07.019