Multi-Task Learning for the Diagnosis of Alzheimer's Disease

• Activity Number: 126
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Guan Yu* and Yufeng Liu and Dinggang Shen
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina and The University of North Carolina at Chapel Hill
• Keywords: Alzheimer's disease ; magnetic resonance imaging ; multi-task learning ; partial correlation ; positron emission tomography

Abstract

Accurate diagnosis of Alzheimer's disease is very important. Over the last decade, various machine learning methods have been proposed to predict disease status and clinical scores from brain images. It is worth noting that many features extracted from brain images are correlated. In this case, feature selection combined with the correlation information among features can improve the prediction performance. Typically, the correlation information among features can be modeled by the connectivity of an undirected graph, where each node represents one feature and each edge indicates that the two involved features are correlated significantly. In this paper, we propose a new multi-task learning method incorporating this undirected graph information to predict multiple response variables jointly. Specifically, we utilize a new latent group Lasso penalty to encourage the correlated features to be selected together. Furthermore, this penalty also encourages the intrinsic correlated tasks to share a common feature subset. Compared with the other methods, our method has very promising performance on the simulated data sets and the Alzheimer's Disease Neuroimaging Initiative (ADNI) data set.