Branch-and-Bound for Model Selection and Its Computational Complexity

被引:24
作者
Thakoor, Ninad [1 ]
Gao, Jean [2 ]
机构
[1] Univ Calif Riverside, Ctr Res Intelligent Syst, Engn Unit 2, EBU2, Riverside, CA 92521 USA
[2] Univ Texas Arlington, Comp Sci & Engn Dept, Arlington, TX 76019 USA
关键词
Clustering; segmentation; combinatorial optimization; branch-and-bound; model selection; SEARCH; SEGMENTATION; ALGORITHMS; MOTION;
D O I
10.1109/TKDE.2010.156
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Branch-and-bound methods are used in various data analysis problems, such as clustering, seriation and feature selection. Classical approaches of branch-and-bound based clustering search through combinations of various partitioning possibilities to optimize a clustering cost. However, these approaches are not practically useful for clustering of image data where the size of data is large. Additionally, the number of clusters is unknown in most of the image data analysis problems. By taking advantage of the spatial coherency of clusters, we formulate an innovative branch-and-bound approach, which solves clustering problem as a model-selection problem. In this generalized approach, cluster parameter candidates are first generated by spatially coherent sampling. A branch-and-bound search is carried out through the candidates to select an optimal subset. This paper formulates this approach and investigates its average computational complexity. Improved clustering quality and robustness to outliers compared to conventional iterative approach are demonstrated with experiments.
引用
收藏
页码:655 / 668
页数:14
相关论文
共 41 条
[1]  
[Anonymous], 1993, Modern Heuristics Technics for Combinatorial Problems: Tabu Search pp
[2]  
[Anonymous], 1996, ISIRR96443 U SO CAL
[3]  
[Anonymous], 2001, Robotica, DOI DOI 10.1017/S0263574700223217
[4]  
[Anonymous], STRUCTURE MOTION TOO
[5]  
[Anonymous], 2012, Optimal Estimation of Dynamic Systems
[6]  
[Anonymous], J MACH LEARN RES
[7]  
[Anonymous], 2011, Pei. data mining concepts and techniques
[8]  
[Anonymous], 2002, COMPUTER VISION MODE, DOI 10.5555/580035
[9]  
Brusco M. J., 2005, BRANCH AND BOUND APP
[10]  
BURNHAM K.P., 2002, MODEL SELECTION MULT, P352