Exclusive feature selection and multi-view learning for Alzheimer’s Disease

Abstract

In Alzheimer’s Disease (AD) studies, high dimension and small sample size have been always an issue and it is common to apply a dimension reduction method to predict the early diagnosis of AD. In this paper, we propose a multi-view feature selection algorithm embedded with exclusive lasso learning and sparse learning. It extracts the feature subsets that best represent the symptoms of patients through feature selection, so as to reduce the dimension and achieve a better diagnosis rate. Firstly, in order to overcome the limitation of non-overlapping, the features under different view are clustered by fuzzy C-means clustering. Then, the exclusive group lasso learning is performed according to the clustering results and each view is sparsely learned through the l2,1-norm, resulting in better removal of redundant features. Finally, the results of each view are combined to obtain the final features subsets. This exclusive lasso learning combined through multiple views is novel in clinical practice and can effectively target AD. At the same time, the experimental results show that our method could achieve better results compared to its competing methods.