Prediction of Dementia Using Whale Optimization Algorithm Based Convolutional Neural Network

Dementia, which is a chronic disorder affects the mental health of the person. Generally, it causes problems in memory disorder, affects thinking abilities of a person and also affects day to day life activities. There is no such treatment that cures the dementia fully. Thus, it is essential to predict the occurrence of Dementia in advance. This paper aims to detect Dementia using Convolutional Neural Network (CNN). The hyperparameters in CNN play a prominent role for the performance of the model. The optimal values of hyperparameters are thus essential in boosting the accuracy of the model. Whale Optimization Algorithm (WOA), a metaheuristic algorithm is used to tune the hyperparameters such as mini batch size and dropout rate in AlexNet, a CNN architecture. The proposed WOA-CNN is compared with Conventional CNN, AlexNet architecture in terms of accuracy and loss. The experimental results validate that the proposed WOA-CNN achieves 98.987% accuracy in detecting Dementia which is 3.46% greater than CNN.