Feature Selection Using Adaptive Weber Distribution Based Flower Pollination Algorithm for Alzheimer’s Disease Classification

Alzheimer’s Disease (AD) is a neurodegenerative disease which affects old people, causing cognitive decline and memory loss. However, the classification accuracy in existing methods is low as the feature selection method is not precise, which results in the consideration of irrelevant features for the classification. In this research, an Adaptive Weber Distribution – Flower Pollination Algorithm (AWD-FPA) method is proposed for feature selection that selects relevant features from whole feature subset by eliminating the irrelevant features. Then, the Gated Recurrent Units (GRU) with Rectified Linear Unit (ReLU) based classification is performed which classifies the AD classes with high classification accuracy. The proposed AWD-FPA and GRU with ReLU method obtains 97.80% accuracy in AD vs CN binary class, 96.31% accuracy in AD vs MCI binary class, 99.62% accuracy in CN vs MCI binary class and 97.21% accuracy in AD vs MCI vs CN multi class which is effective than existing methods.