Smartwatch-Based Legitimate User Identification for Cloud-Based Secure Services

Table 2

Confusion matrix according to the users using 400 samples per window with accelerometer and gyroscope sensors.

Users	Accelerometer sensor						Gyroscope sensor
	User 1	User 2	User 3	User 4	User 5	User 6	User 1	User 2	User 3	User 4	User 5	User 6
DT classifier
User 1	1710	2	6	40	1	42	1246	1	0	12	3	44
User 2	0	1796	48	9	35	6	29	1889	42	1	78	17
User 3	0	40	1774	31	125	0	1	19	1772	30	123	15
User 4	12	13	11	1695	18	45	32	7	19	1699	7	31
User 5	18	24	40	2	1457	0	0	21	69	0	1476	4
User 6	121	16	3	49	1	1880	140	2	4	30	0	1895
SVM classifier
User 1	1473	16	27	76	28	135	1511	32	9	63	1	100
User 2	9	1597	91	40	84	9	53	1787	51	10	110	40
User 3	14	165	1568	51	163	0	18	36	1576	26	181	20
User 4	26	81	75	1644	22	68	69	3	37	1627	1	33
User 5	50	48	134	9	1346	0	4	39	119	0	1344	29
User 6	152	0	8	53	16	1781	185	4	4	69	5	1862
KNN classifier
User 1	1514	11	15	59	19	113	1428	13	18	17	0	79
User 2	1	1679	131	24	132	23	52	1779	109	29	238	88
User 3	10	113	1616	12	111	7	0	12	1638	37	184	36
User 4	41	56	11	1575	5	83	97	16	27	1680	8	87
User 5	15	46	89	3	1332	0	4	37	53	21	1282	3
User 6	238	13	10	68	7	1888	183	4	6	43	0	1749
NB classifier
User 1	1008	49	108	178	24	205	593	83	4	12	62	102
User 2	1	1370	112	35	169	2	79	1264	28	12	244	68
User 3	40	230	1069	32	218	35	131	94	974	191	96	44
User 4	221	211	157	1079	29	127	364	50	561	1104	601	106
User 5	190	83	350	28	1472	119	74	177	193	41	611	23
User 6	338	10	58	438	52	1566	487	202	132	485	82	1693