Daily Human Physical Activity Recognition Based on Kernel Discriminant Analysis and Extreme Learning Machine

Table 1

The confusion matrices between the 6 classes of activities. R indicates running, ST indicates sitting, SD indicates standing, W indicates walking, U indicates going upstairs, and D indicates going downstairs.

(a) ELM based on original features

	R	ST	SD	W	U	D
R	1	0	0	0	0	0
ST	0	1	0	0	0	0
SD	0	0	0.999	0	0	0.001
W	0	0	0	0.985	0.005	0.01
U	0	0	0	0.015	0.968	0.017
D	0	0	0	0.021	0.009	0.97

(b) ELM based on KDA features

	R	ST	SD	W	U	D
R	1	0	0	0	0	0
ST	0	1	0	0	0	0
SD	0	0	1	0	0	0
W	0	0	0	1	0	0
U	0	0	0	0	0.999	0.001
D	0	0	0	0.001	0	0.999

(c) LM-BP based on original features

	R	ST	SD	W	U	D
R	0.998	0	0	0.002	0	0
ST	0	0.998	0	0	0.002	0
SD	0	0	0.996	0.003	0	0.001
W	0	0	0	0.98	0.006	0.014
U	0	0	0	0.018	0.962	0.02
D	0	0	0.006	0.024	0.012	0.958

(d) LM-BP based on KDA features

	R	ST	SD	W	U	D
R	1	0	0	0	0	0
ST	0	1	0	0	0	0
SD	0	0	1	0	0	0
W	0	0	0	0.99	0.006	0.004
U	0	0	0	0.01	0.988	0.002
D	0	0	0	0.009	0.006	0.985

(e) LS-SVM based on original features

	R	ST	SD	W	U	D
R	0.993	0	0	0.006	0	0.001
ST	0	0.996	0	0	0.004	0
SD	0	0	0.997	0.001	0	0.002
W	0	0	0	0.981	0.006	0.013
U	0	0	0	0.01	0.976	0.014
D	0	0	0	0.02	0.012	0.968

(f) LS-SVM based on KDA features

	R	ST	SD	W	U	D
R	1	0	0	0	0	0
ST	0	1	0	0	0	0
SD	0	0	1	0	0	0
W	0	0	0	0.998	0	0.002
U	0	0	0	0.003	0.993	0.004
D	0	0	0	0.002	0.002	0.996