Knowledge Mining from Clinical Datasets Using Rough Sets and Backpropagation Neural Network

Table 5

Parameters of BPNN.

Number of layers	Input layer: 1 with 9–13 features for hepatitis; 9 features for breast cancer; 3–7 features for heart disease
	Hidden layer: 1 with 25 nodes ()
	Output layer: 1 with class label (0 or 1)
Activation function	Hidden layer: tangent sigmoid
	Output layer: linear
Learning algorithm	Backpropagation
Dataset division	Random division