Multi-layer multi-view topic model for classifying advertising video

作者：

Highlights：

• We proposed a new video representation from the perspective of multiple-view fusion, by using multi-layer multi-view LDA framework.

• The proposed video representation scheme solved the problem of ads video classification effectively.

• We built a publicly advertisement video dataset, which can be shared in the research of video advertising.

摘要

•We proposed a new video representation from the perspective of multiple-view fusion, by using multi-layer multi-view LDA framework.•The proposed video representation scheme solved the problem of ads video classification effectively.•We built a publicly advertisement video dataset, which can be shared in the research of video advertising.

论文关键词：Video representation,Ad video classification,Multi-layer,Multi-view,Topic model

论文评审过程：Received 9 August 2016, Revised 30 November 2016, Accepted 1 March 2017, Available online 2 March 2017, Version of Record 12 March 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.03.003