基于划分自适应融合的多视角模糊聚类算法

针对多视角聚类任务如何更好地实现视角间的合作之挑战, 提出一种新的视角融合策略. 该策略首先为每个视角设置一个划分, 然后通过自适应学习获取一个融合权重矩阵对每个视角的划分进行自适应融合, 最终利用视角集成方法得到全局划分结果. 将上述策略应用到经典的FCM(Fuzzy ??-means) 模糊聚类框架, 提出相应的多视角模糊聚类算法. 在模拟数据集和UCI 数据集上的实验结果均显示, 所提出的算法较几种相关聚类算法在应对多视角聚类任务时具有更好的适应性和更好的聚类性能.

Multi-view fuzzy clustering algorithm based on partition adaptive-fusion

How to effectively realize the view fusion in the multi-view clustering is an important challenge. A new view fusion strategy is proposed for this task. Firstly, the partition matrix of each view is setup. And then the adaptive-fusion in each view partition is made based on the multiple view partitions by an adaptive-fusion weighting matrix. Finally, the global partition is obtained by using an integration approach. This strategy is integrated with the classic fuzzy ??-means clustering framework, and the corresponding multi-view clustering algorithm is presented. Experimental studies are carried out on the synthetic and UCI real-world multi-view datasets. The experimental results show that the proposed algorithm outperforms several existing related algorithms in the adaptative abilities and clustering performance.