Font Size: 

Multi-view clustering is a data mining task in which a data set is processed by several algorithms observing different features of the same data. The main difficulty of this task is to detect whether or not sharing informations between the views may be beneficial: Some views contain mostly noisy features, while others simply contain features which lead to different clusters. One of the challenges of multi-view clutering is therefore to find which views should work together or not. Within this context, in this article we propose an optimisation method which sets the exchange weights between the different algorithms based on the maximization of the global likelihood function.