Graph-Guided Universum Learning in Generalized Eigenvalue Proximal SVMs for Alzheimer's Disease Classification
Researchers have developed two novel graph-guided learning models, UG-GEPSVM and IUG-GEPSVM, to improve the classification of Alzheimer's disease using structural MRI data. These models leverage mild cognitive impairment (MCI) subjects as 'Universum' data, constructing a graph to capture the geometric relationships among them. This approach enhances the learning process by incorporating graph-based regularization, outperforming existing methods with an average AUC of 88.07% on the ADNI dataset. AI
IMPACT These models offer a more nuanced approach to medical image analysis, potentially improving diagnostic accuracy for neurodegenerative diseases.