Table 2. The prediction results of candidate classifiers generated from the final PSO (pbest) solutions and randomly generated solutions on the testing dataset using neural network classifiers with two processing layers for the 5-fold cross-validation on the MNIST dataset.

(Here, Avg denotes the average prediction result across all candidate classifiers, and N of c denotes the number of combined classifiers).

	DNN-PSO					DNN-NONPSO
N of c	Fold-1	Fold-2	Fold-3	Fold-5	Fod-5	Fold-1	Fold-2	Fold-3	Fold-4	Fold-4
1	0.9785	0.9801	0.9770	0.9746	0.9808	0.9774	0.9708	0.9738	0.9750	0.1121
2	0.9698	0.9804	0.9788	0.9705	0.9775	0.9714	0.9674	0.9779	0.9733	0.9754
3	0.9774	0.9791	0.9807	0.9726	0.9791	0.9797	0.9348	0.9764	0.9764	0.9786
4	0.9816	0.9794	0.9764	0.9779	0.9831	0.9755	0.9787	0.9769	0.9741	0.9798
5	0.9795	0.9797	0.9781	0.9764	0.9779	0.9809	0.9768	0.9790	0.9720	0.9761
6	0.9791	0.9802	0.9736	0.9775	0.9784	0.9794	0.9754	0.9776	0.9745	0.9768
7	0.9799	0.9761	0.9784	0.9728	0.9786	0.9782	0.9766	0.9768	0.9663	0.9823
8	0.9807	0.9762	0.9692	0.9692	0.9804	0.9759	0.9771	0.9761	0.9551	0.9776
9	0.9795	0.9776	0.9756	0.9736	0.9809	0.9816	0.9769	0.9731	0.9652	0.9794
10	0.9759	0.9804	0.9776	0.9781	0.9785	0.9776	0.9792	0.9768	0.9741	0.9771
11	0.9805	0.9801	0.9779	0.9740	0.9792	0.9802	0.9761	0.9735	0.9735	0.9814
12	0.9751	0.9801	0.9756	0.9732	0.9763	0.9767	0.9816	0.9754	0.9755	0.9784
13	0.9817	0.9789	0.9761	0.9758	0.9739	0.9784	0.9774	0.9773	0.9741	0.9790
14	0.9745	0.9781	0.9781	0.9759	0.9809	0.9711	0.9735	0.9719	0.9721	0.9787
15	0.9797	0.9798	0.9776	0.9761	0.9826	0.9784	0.9797	0.9660	0.9755	0.9783
16	0.9789	0.9782	0.9769	0.9748	0.9821	0.9747	0.9797	0.9651	0.9731	0.9797
17	0.9761	0.9801	0.9786	0.9741	0.9815	0.9744	0.9531	0.9781	0.9749	0.9376
18	0.9751	0.9772	0.9787	0.9729	0.9789	0.9795	0.9739	0.9779	0.9685	0.9802
19	0.9792	0.9794	0.9774	0.9767	0.9807	0.9774	0.9766	0.9749	0.9749	0.9577
20	0.9765	0.9782	0.9763	0.9726	0.9749	0.9791	0.9664	0.9601	0.9734	0.9784
Avg	0.9780	0.9790	0.9769	0.9745	0.9793	0.9774	0.9726	0.9742	0.9721	0.9322
Std	0.0029	0.0013	0.0023	0.0023	0.0024	0.0029	0.0110	0.0050	0.0050	0.1933