Fabric Defect Detection Using Local Homogeneity Analysis and Neural Network

Table 2

Training parameters.

Parameter	Value
Number of inputs/outputs	4/1
Number of hidden layers	1
Number of hidden layer neurons	9
Learning coefficient for hidden layer	0.6
Activation function	Bipolar Sigmoid
Initial weight values (random)	0~0.1
Learning rule delta	Delta
Convergence error
Number of iterations	50120