Table 6: The confusion matrix for the Bayesian model averaging of artificial neural networks and the two class problem. The last but one line lists the total number of light curves (TOT) to define every class. The last line lists the correct classification rate (CC) for every class separately as measured by 10-fold cross validation. Separation between: A: eclipsing binaries (ECL) and all other types; B: Cepheids (CEP) and all other types; C: long period variables (LPV) and all other types except ECL and CEP; D: RR Lyrae stars (RR) from all other types except ECL, CEP and LPV.

A			B			C			D
	ECL	REST		CEP	REST		LPV	REST		RR	REST
ECL	374	2	CEP	302	18	LPV	164	18	RR	205	8
REST	1	1377	REST	25	1034	REST	23	847	REST	10	642
TOT	375	1379	TOT	327	1052	TOT	187	865	TOT	215	650
CC (%)	99.73	99.85	CC (%)	92.35	98.28	CC (%)	87.70	97.91	CC (%)	95.34	98.76