Bagging for Path-Based Clustering Bernd Fischer, Student Member, IEEE, and.pdf

下载文档

4
0
约2.97万字
约 6页
2017-04-09 发布于江苏
举报
版权申诉
保障服务

Bagging for Path-Based Clustering Bernd Fischer, Student Member, IEEE, and.pdf

1、本文档共6页，可阅读全部内容。
2、有哪些信誉好的足球投注网站（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

Bagging for Path-Based Clustering Bernd Fischer, Student Member, IEEE, and

Bagging for Path-Based Clustering Bernd Fischer, Student Member, IEEE, and Joachim M. Buhmann, Member, IEEE Abstract—A resampling scheme for clustering with similarity to bootstrap aggregation (bagging) is presented. Bagging is used to improve the quality of path- based clustering, a data clustering method that can extract elongated structures from data in a noise robust way. The results of an agglomerative optimization method are influenced by small fluctuations of the input data. To increase the reliability of clustering solutions, a stochastic resampling method is developed to infer consensus clusters. A related reliability measure allows us to estimate the number of clusters, based on the stability of an optimized cluster solution under resampling. The quality of path-based clustering with resampling is evaluated on a large image dataset of human segmentations. Index Terms—Clustering, resampling, color segmentation. 1 INTRODUCTION CLUSTERING objects into separated groups is an important topic in exploratory data analysis and pattern recognition. Many clustering techniques group the data objects together to “compact” clusters with the explicit or implicit assumption that all objects within one group are either mutually similar to each other or they are similar with respect to a common representative or centroid. The most prominent example of this concept is k-means clustering for vectorial data. Pairwise clustering [1] and distributional clustering [2], [3] are analogous examples for proximity data and histogram data. In many data analysis and pattern recognition scenarios, however, similarity between two objects is not only established by direct comparison, but it can be induced by mediating objects in between, i.e., two objects might be considered similar when they are connected by a chain of intermediate objects where all dissimilarities or distances between neighboring objects in the chain are small. The path-based clustering approach [4] is able to extract