A Joint Optimization of Momentum Item and Levenberg-Marquardt Algorithm to Level Up the BPNN’s Generalization Ability

Table 1

The recognition results of different training methods for Fisher Iris data.

Real labels	Preprocessed output			Postprocessed output
	ELM-BPNN	NLM-BPNN	Proposed LM-BPNN	ELM-BPNN	NLM-BPNN	Proposed LM-BPNN
1	0.99999933	0.99999897	0.99999953	1	1	1
1	0.99999930	0.99999965	0.99999953	1	1	1
1	0.99999937	0.99999897	0.99999953	1	1	1
1	0.99999933	0.99999897	0.99999953	1	1	1
1	0.99999933	0.99999897	0.99999953	1	1	1
2	1.99999855	31.3832497	2.00000393	2	31	2
2	2.00000235	2.51428256	2.00000169	2	3	2
2	1.99996337	−0.8690897	2.00210956	2	−1	2
2	2.00000217	1.99807531	2.00001190	2	2	2
2	2.00000623	1.80912596	2.00193468	2	2	2
3	2.99999644	2.99999999	3.00000059	3	3	3
3	2.49818368	1.99998783	2.50605341	2	2	3
3	2.99037581	2.99988701	2.99999197	3	3	3
3	3.00000291	2.99999999	3.00000059	3	3	3
3	2.99662415	2.99542888	2.79537363	3	3	3