Fig. 3
From: Machine learning and statistics shape a novel path in archaeal promoter annotation
ROC curves for the best ANN simulation in each archaeon. Once the best architecture for classifying archaea (each organism is presented in a separate panel) with neural networks was defined, the classification threshold was adjusted to produce ROC curves. The default output neuron yields a value and if it's bigger than 0.5, it gets classified as a promoter, otherwise, it is classified as a non-promoter. Each tick in the ROC curves represents an adjusted decision threshold, varying from 0 (x axis = 0, y axis = 0) to 1 (x axis = 1, y axis = 1)