Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery

Bures-Wasserstein Barycenters and Low-Rank Matrix Recovery
复制标题

DOI:
--
复制
发表时间:
2022-10
期刊:
--
影响因子:
--
通讯作者:
Tyler Maunu;Thibaut Le Gouic;P. Rigollet
Tyler Maunu;Thibaut Le Gouic;P. Rigollet
中科院分区:
其他
文献类型:
--
作者:
Tyler Maunu;Thibaut Le Gouic;P. Rigollet

文献摘要

相似文献

我们重新审视的问题,恢复一个低秩半正定矩阵的秩一投影使用工具,从最佳运输。更具体地说,我们表明,这个问题的变分公式相当于计算Wasserstein重心。反过来,这种新的视角,使新的几何一阶方法的发展具有强大的收敛保证Bures-Wasserstein距离。模拟数据的实验表明,我们的新方法比现有的方法的优势。
We revisit the problem of recovering a low-rank positive semidefinite matrix from rank-one projections using tools from optimal transport. More specifically, we show that a variational formulation of this problem is equivalent to computing a Wasserstein barycenter. In turn, this new perspective enables the development of new geometric first-order methods with strong convergence guarantees in Bures-Wasserstein distance. Experiments on simulated data demonstrate the advantages of our new methodology over existing methods.