GIFSL - grafting based improved few-shot learning

作者：

Highlights：

• A novel approach to improve few-shot learning.

• Filter-grafting improves the representational capacity of the network.

• Self-supervision and knowledge distillation improve the discriminative power.

• Also improves the cross-domain few-shot learning performance.

摘要

•A novel approach to improve few-shot learning.•Filter-grafting improves the representational capacity of the network.•Self-supervision and knowledge distillation improve the discriminative power.•Also improves the cross-domain few-shot learning performance.

论文关键词：Few-shot learning,Grafting,Self-supervision,Distillation,Deep learning,Object recognition

论文评审过程：Received 7 June 2020, Revised 12 August 2020, Accepted 13 August 2020, Available online 19 August 2020, Version of Record 26 August 2020.

论文官网地址：https://doi.org/10.1016/j.imavis.2020.104006