Skip to main content
Fig. 5 | BMC Bioinformatics

Fig. 5

From: Implementation of ensemble machine learning algorithms on exome datasets for predicting early diagnosis of cancers

Fig. 5

Train-validation accuracy versus epochs and train-validation loss versus epochs for neural network with SMOTE oversampling. From the graph, it can be seen that the validation accuracy stalls around 40 epochs and has only slight variation after that hence training for 40 epochs should be sufficient to provide same performance as training for 100 epochs. From the validation loss graph, it is noted that after around 50 epochs the model starts to overfit for the training data and hence stopping it after that should prevent it from overfitting

Back to article page