Research on Dynamic Programming Strategy of Bayesian Network Structure Learning

Table 4

Time-consuming comparison between three algorithms under standard network data sets (s).

BNs	Nodes	Edges	Tns	Sample sizes	DFSDP	CBDP	BFSDP
Weather	4	4	8	1000	0.1935	0.0590	0.1742
				2000	0.2016	0.0899	0.1800
				5000	0.2120	0.1593	0.1865
				10000	0.2326	0.2240	0.1948
Cancer	5	4	10	1000	0.3348	0.0511	0.2391
				2000	0.3620	0.1077	0.2487
				5000	0.3768	0.1854	0.2693
				10000	0.4308	0.3494	0.2905
Earthquake	5	4	10	1000	0.3331	0.0849	0.2470
				2000	0.3493	0.1122	0.2487
				5000	0.3749	0.1615	0.2673
				10000	0.4234	0.2642	0.2981
Survey	6	6	14	1000	0.7828	0.1668	0.3391
				2000	0.7871	0.2790	0.3882
				5000	0.8565	0.5817	0.4355
				10000	1.0047	1.0419	0.4961
Asia	8	8	16	1000	7.8807	1.6342	1.3848
				2000	8.4305	1.7328	1.5450
				5000	8.5538	2.1903	1.8037
				10000	9.2984	2.9253	2.1439
Sachs	11	17	33	1000	1.02e + 003	464.0321	14.55
				2000	1.03e + 003	289.0895	20.38
				5000	1.05e + 003	238.4937	27.91
				10000	1.07e + 003	230.4499	38.11
Child	20	25	60	1000	OT	OT	2.1348e + 005
				2000			2.5035e + 005
				5000			OT
				10000			OT
Insurance	27	52	89	1000	OM	OM	OM
				2000
				5000
				10000