Diabetes is a disease caused by high sugar levels. Currently, diabetes is one of the most common diseases in the number of people with diabetes worldwide. The increase in diabetes is caused by the delay in establishing the diagnosis of the disease. Therefore, an initial action is needed as a solution that requires the most appropriate and accurate data mining to manage diabetes mellitus. The algorithms used are artificial neural network algorithms, namely Restricted Boltzmann Machine and Backpropagation. This research aims to compare the two algorithms to find which algorithm can produce high accuracy, and determine which algorithm is more accurate in detecting diabetes mellitus. Several stages were involved in this research, including data collection, data pre-processing, data processing, and evaluation models. This research shows that the Restricted Boltzmann Machine algorithm achieved accuracy of 82.02% while the Backpropagation algorithm reached87.01% when using the normalization method. Thus, the diabetes mellitus dataset used can be said to have a better value for the backpropagation algorithm than the restricted Boltzmann machine algorithm.
