Incomplete multiview nonnegative representation learning with multiple graphs

作者：

Highlights：

• We build a novel incomplete multiview nonnegative representation learning framework, referred to as IMNRL.

• IMNRL learns a consensus nonnegative representation and view-specific representations simultaneously.

• The nonnegative representation retains the graph information on all views, and it reveals the clustering results.

• IMNRL achieves state-of-the-art incomplete multiview clustering results on different incomplete cases.

摘要

•We build a novel incomplete multiview nonnegative representation learning framework, referred to as IMNRL.•IMNRL learns a consensus nonnegative representation and view-specific representations simultaneously.•The nonnegative representation retains the graph information on all views, and it reveals the clustering results.•IMNRL achieves state-of-the-art incomplete multiview clustering results on different incomplete cases.

论文关键词：Multiview clustering,Graph learning,Incomplete multiview clustering,Nonnegative matrix factorization

论文评审过程：Received 23 January 2021, Revised 19 September 2021, Accepted 30 October 2021, Available online 1 November 2021, Version of Record 15 November 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108412