Table 5.

The SHD between the structure learned by the proposed method and the I-map with the fewest parameters among all the ANB structures and the KLD between the learned class variable posterior by the proposed method and learned one using the true structure.

Network	Variables	Sample Size	SHD-(Proposal, I-Map ANB)	KLD-(Proposal, True Structure)
		100	3	$2.31\times {10}^{-2}$
		500	2	$1.24\times {10}^{-1}$
		1000	2	$7.63\times {10}^{-2}$
ASIA	8	5000	1	$3.67\times {10}^{-3}$
		10,000	0	$9.26\times {10}^{-4}$
		50,000	0	$6.28\times {10}^{-4}$
		100,000	0	$3.59\times {10}^{-5}$
		100	1	$8.79\times {10}^{-2}$
		500	1	$2.43\times {10}^{-3}$
		1000	0	0.00
CANCER	5	5000	0	0.00
		10,000	0	0.00
		50,000	0	0.00
		100,000	0	0.00