基于隐空间的低秩稀疏子空间聚类

Low-rank sparse subspace clustering in latent space

提出了一种基于隐空间的低秩稀疏子空间聚类算法,在聚类的过程中可以对高维数据进行降维,同时在低维空间中利用稀疏表示和低秩表示对数据进行聚类,大大降低了算法的时间复杂度.在运动分割和人脸聚类问题上的实验证明了算法的有效性.

This paper proposed a novel algorithm named low‐rank sparse subspace clustering in latent space （LatLRSSC ） , it can reduce the dimension and cluster the data lying in a union of subspaces simultaneously . The main advatages of our method is that it is computationally efficient . The effectiveness of the algorithm is demonstrated through experiments on motion segmentation and face clustering .