A New Optimized GA-RBF Neural Network Algorithm

Table 2

The comparison of the performance of each algorithm for wine data set.

Neural networks algorithm	Traditional RBF	GA-RBF	GA-RBF-L
Training success rate, %	90	100	100
Training error	0.37	0.53	0.32
Test error	0.89	1.06	0.67
Number of hidden neurons	14	8	8
Operation time, s	0.27	1.53
Classification accuracy, %	90.94	86.88	96.35