A Novel User Classification Method for Femtocell Network by Using Affinity Propagation Algorithm and Artificial Neural Network

Table 2

Number of samples		ANN training performance		AP + ANN training performance			Performance comparison after AP
Indoor	Outdoor	Training time (sec.)	Number of epochs*	AP clustering time (sec.)	Total training time (sec.)	Number of epochs*	Training time decreases (%)	Number of epochs decreases (%)
5	10	1.5165	21	0.2885	0.7647	13	49.57	38.09
6	11	1.5655	21	0.3633	0.8257	14	47.25	33.33
7	12	1.6221	22	0.3921	0.9071	14	44.07	36.36
8	13	1.6416	23	0.4172	0.934	15	43.10	34.78
9	14	1.6443	24	0.4226	0.9504	16	42.20	33.33
10	15	1.6504	26	0.4244	0.9611	16	41.76	38.46
11	16	1.6541	26	0.439	0.9803	17	40.73	34.61
12	17	1.6669	27	0.4461	0.9848	18	40.92	33.33
13	18	1.6709	27	0.4556	0.9951	19	40.44	29.62
14	19	1.6801	28	0.4671	0.9951	19	40.77	32.14
15	20	1.6819	28	0.4702	0.9964	19	40.75	32.14
16	21	1.688	29	0.4704	0.9982	19	40.86	34.48
17	22	1.7322	30	0.4997	1.0017	21	42.17	30
18	23	1.7767	31	0.5073	1.0397	22	41.48	29.03
19	24	1.7966	32	0.5234	1.0695	22	40.47	31.25
20	25	1.7996	34	0.5484	1.0511	22	41.59	35.29

∗The fraction values of the epochs are expressed by the nearest integer values.