Applying Artificial Neural Network to Predict Semiconductor Machine Outliers

Table 1

Network training and performance tests.

Hidden neuron	1 month (a)		3 months (b)		6 months (c)
	MSE	Train-R	MSE	Train-R	MSE	Train-R
1	0.0162	0.5873	0.0132	0.6773	0.0173	0.5643
2	0.0154	0.6032	0.0136	0.6832	0.0163	0.5932
3	0.0165	0.6244	0.0142	0.6834	0.0166	0.4934
4	0.0166	0.5873	0.0136	0.6583	0.0169	0.5274
5	0.0169	0.5972	0.0139	0.6572	0.0162	0.5843
6	0.0157	0.5878	0.0147	0.6978	0.0164	0.5032
7	0.0168	0.5983	0.0148	0.5783	0.0177	0.5836
8	0.0163	0.5973	0.0143	0.6373	0.0179	0.5787
9	0.0168	0.6036	0.0138	0.6439	0.0185	0.5897
10	0.0169	0.6092	0.0145	0.6899	0.0187	0.5554
Average	0.01641	0.59956	0.01406	0.66066	0.01725	0.55732

Denote: MSE is mean square error; Train-R is correlation coefficient.