proximity of multi-view consensus Clustering Learning

PROJECT TITLE :

Multi-view Consensus Proximity Learning for Clustering

ABSTRACT:

The majority of proximity-based multi-view clustering methods are sensitive to the initial proximity matrix. As a result, the clustering performance is quite unstable when using a variety of different initial proximity matrices. The initial value sensitivity problem is another name for this issue. It is not possible to tune the initial proximity matrix because clustering is an unsupervised learning task. Therefore, a significant but unresolved problem in proximity-based multi-view clustering is the question of how to get around the issue of the initial value being sensitive. This article proposes a brand new multi-view proximity learning method, which is called multi-view consensus proximity learning, in order to accomplish this goal (MCPL). On the one hand, the MCPL method learns the consensus proximity matrix to directly reflect the clustering result by integrating the information of all views in a self-weighted manner and giving a rank constraint on the Laplacian matrix. This allows the method to directly reflect the clustering result. On the other hand, unlike the majority of multi-view proximity learning methods, the proposed MCPL method adopts the data representatives rather than the original data objects in order to learn the consensus proximity matrix. This is a significant departure from the majority of multi-view proximity learning methods. As part of the process of proximity learning, the data representatives will be brought up to date, with the goal of reducing the effect that the initial value has on the effectiveness of the clustering. For the purpose of demonstrating that the proposed method is effective, a large number of experiments are carried out.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here