A SVM-based model-transferring method for heterogeneous domain adaptation

作者：

Highlights：

• We propose a new SVM-based model-transferring method for adaptation.

• Our method applies adaptation in the one-dimensional discrimination space.

• The proposed method can handle heterogeneous domains.

• Based on proposed model-transferring method, we design a new metric for measuring the adaptability between two domains.

摘要

•We propose a new SVM-based model-transferring method for adaptation.•Our method applies adaptation in the one-dimensional discrimination space.•The proposed method can handle heterogeneous domains.•Based on proposed model-transferring method, we design a new metric for measuring the adaptability between two domains.

论文关键词：SVM-based method,Model-transferring method,Heterogeneous domain adaptation,Object detection,Image classification

论文评审过程：Received 19 July 2015, Revised 26 February 2016, Accepted 3 March 2016, Available online 11 March 2016, Version of Record 12 April 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.03.009