Alzheimer's disease (AD) is a leading cause of dementia and other forms of cognitive decline in the elderly and middle-aged, and the rising prevalence of the illness will place a strain on healthcare systems. Due to its rapidly ageing population, China now has more Alzheimer's disease patients than any other country. Therefore, it is of paramount importance to find a way to make an early and accurate diagnosis of Alzheimer's disease and to take effective action. Also over the last few decades, many automated technologies and approaches have been created for the diagnosis of Alzheimer's disease (AD). In order to mitigate the condition's effect on the patient's mental health, there are diagnostic strategies that prioritise speed, precision, and early detection machine learning and deep learning have significantly improved the diagnostic performance of medical imaging systems for Alzheimer's disease. However, the existence of highly correlated anatomical features of the brain poses a significant challenge to multi-class categorization. Nevertheless, the vast majority of deep learning models fail to deliver acceptable results in real-world situations. So in order to tackle this situation, we have built an ensemble classifier to improve Deep CNN's learning outcomes by fusing two models, CNN and VGG-16, into a single model. Based on OASIS dataset, we've gathered 8,980 MRI images to test our suggested technique. According to our research, we have observed that an ensemble based model (CNN+VGG-16 Net) outperforms individual deep learning models (Traditional CNN and VGG-16 Net).