多视角模糊双加权可能性聚类算法

为解决传统可能性聚类算法（PCM）无法满足多视角学习场景聚类的实际问题，并进一步考虑到现有多视角聚类算法尚未重视的视角权重及视角内特征权重优化问题，本文提出一种新的具备最佳视角及最优特征划分能力的多视角模糊双加权可能性聚类算法（MV-FDW-PCM）。该算法将基于传统的PCM算法，给出了详细的多视角聚类学习框架使得PCM算法具备多视角聚类能力，进而通过引入视角间模糊加权机制及视角内属性模糊加权机制解决视角间权重及视角内特征权重优化问题。实验结果表明，所提的MV-FDW-PCM算法在面对多视角聚类问题时较以往算法具有更佳的聚类效果。

Multi-view fuzzy double-weighting possibility clustering algorithm

To solve the problem that traditional possibility clustering algorithms (PCM) barely achieve multi-view clustering, and considering that the optimization of views and feature weights has not been regarded as important in existing multi-view clustering algorithms, this paper proposes a new multi-view fuzzy double-weighted possibility clustering algorithm (MV-FDW-PCM). The algorithm is based on the traditional PCM algorithm, and it gives a detailed multi-view clustering learning framework, which gives it its own multi-view clustering ability. It realizes the optimization of the weight of view and the feature weight within the view by the introduction of an inter-view fuzzy weighting mechanism and an inside-view attribute fuzzy weighting mechanism. The experimental results show that the proposed MV-FDW-PCM algorithm has better clustering performance than the previous algorithms regarding multi-view clustering.