Fundamental sampling patterns for low-rank multi-view data completion

作者：

Highlights：

• A geometrical sampling pattern is obtained that guarantees the retrieval of the low-rank multi-view data in certain ways.

• An analysis is proposed that incorporates several rank constraints of the multi-view data simultaneously.

• Combinatorial analysis is proposed to find the sampling rate that guarantees the obtained geometrical sampling pattern hold with high probability.

摘要

•A geometrical sampling pattern is obtained that guarantees the retrieval of the low-rank multi-view data in certain ways.•An analysis is proposed that incorporates several rank constraints of the multi-view data simultaneously.•Combinatorial analysis is proposed to find the sampling rate that guarantees the obtained geometrical sampling pattern hold with high probability.

论文关键词：Multi-view learning,Low-rank matrix completion,Sampling pattern,Sampling rate,Non-convex optimization,Rank decomposition

论文评审过程：Received 8 April 2019, Revised 16 December 2019, Accepted 24 February 2020, Available online 29 February 2020, Version of Record 6 March 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107307