A Graph-Based Projection Approach for Semi-supervised Clustering
A Graph-Based Projection Approach for Semi-supervised Clustering
复制标题
DOI：
10.1007/978-3-642-15037-1_1
复制
发表时间：
2010-08
期刊：
影响因子：
--
通讯作者：
Tetsuya Yoshida;Kazuhiro Okatani
中科院分区：
文献类型：
--
作者：
Tetsuya Yoshida;Kazuhiro Okatani
提出了一种基于图投影的半监督聚类方法。在我们的方法中，整个数据被表示为一个边加权图与实例之间的成对相似性。图表示使得能够处理两种成对约束以及同一统一表示上的成对相似性。然后，为了反映成对约束对聚类过程的影响，通过图论中的收缩和谱图理论中的图拉普拉斯算子对图进行了修改。通过利用约束以及实例之间的相似性，整个数据通过修改后的图投影到一个子空间，并在投影表示进行数据聚类。在几个真实的世界数据集上对所提出的方法进行了评估。结果是令人鼓舞的，并表明所提出的方法的有效性。
This paper proposes a graph-based projection approach for semi-supervised clustering based on pairwise relations among instances. In our approach, the entire data is represented as an edge-weighted graph with the pairwise similarities among instances. Graph representation enables to deal with two kinds of pairwise constraints as well as pairwise similarities over the same unified representation. Then, in order to reflect the pairwise constraints on the clustering process, the graph is modified by contraction in graph theory and graph Laplacian in spectral graph theory. By exploiting the constraints as well as similarities among instances, the entire data are projected onto a subspace via the modified graph, and data clustering is conducted over the projected representation. The proposed approach is evaluated over several real world datasets. The results are encouraging and indicate the effectiveness of the proposed approach.