
For FullText PDF, please login, if you are a member of IEICE,
or go to Pay Per View on menu list, if you are a nonmember of IEICE.

Energy Minimization over mBranched Enumeration for Generalized Linear Subspace Clustering
Chao ZHANG
Publication
IEICE TRANSACTIONS on Information and Systems
Vol.E102D
No.12
pp.24852492 Publication Date: 2019/12/01
Online ISSN: 17451361
DOI: 10.1587/transinf.2019EDP7138
Type of Manuscript: PAPER Category: Artificial Intelligence, Data Mining Keyword: general subspace clustering, energy minimization, multibranch enumeration,
Full Text: FreePDF(832KB)
Summary:
In this paper, we consider the clustering problem of independent general subspaces. That is, with given data points lay near or on the union of independent lowdimensional linear subspaces, we aim to recover the subspaces and assign the corresponding label to each data point. To settle this problem, we take advantages of both greedy strategy and energy minimization strategy to propose a simple yet effective algorithm based on the assumption that an mbranched (i.e., perfect mary) tree which is constructed by collecting mnearest neighbor points in each node has a high probability of containing the nearexact subspace. Specifically, at first, subspace candidates are enumerated by multiple mbranched trees. Each tree starts with a data point and grows by collecting nearest neighbors in the breadthfirst search order. Then, subspace proposals are further selected from the enumeration to initialize the energy minimization algorithm. Eventually, both the proposals and the labeling result are finalized by iterative reestimation and labeling. Experiments with both synthetic and realworld data show that the proposed method can outperform stateoftheart methods and is practical in real application.

