<![CDATA[Stroke is one of a high mortality disease in Indonesia. A various ways can be done to detect stroke, such as blood test. The result is known just after a few hour. Unfortunately, in some case it took a long time to find out whether a patient at risk of stroke or not. The level of risk can be easily done with a system. Multi-layer perceptron (MLP) network is one of artificial neural network (ANN) model which has a random weight from backpropagation (BP) learning. This study is doing optimization to obtain proper weights, using genetic algorithm (GA) as a training method, so that the classification results are more accurate. Implementation, testing, and analysis are done in BP learning algorithm and GA to compare its accuracy on classifying the risk level of stroke. MSE value obtained in testing phase is 0.0122 with number of iteration = 190, number of neuron in hidden layer = 10, and learning rate = 0.9. While in testing phase of GA obtained 0.0549 with population size = 100, generation size = 400, Cr = 0.8, and Mr = 0.2. In final result, average data accuracy of BP is 88.40% with average MSE value is 0.0122 and GA is 60.60% with average MSE value 0.0549 by 10 times trial.]]>
Copyrights © 2018
