Table 2. Averaged classification accuracies (for initial dip) of all subjects for all features: mean, slope, minimum (min), kurtosis (kurt), skewness (skew).

Feature sets		Window size (sec)	Classification accuracies (%)
			1 s	3 s	5 s	7 s	10 s	15 s
1	mean - slope	0 ∼ 1	54.6	56.7	71.6	61.7	64.3	61.3
		0 ∼ 2.5	56.7	56.9	73.3	62.2	63.7	61.9
		0 ∼ 4	56.8	56.5	74.0	61.5	63.5	61.5
2	mean - min	0 ∼ 1	55.5	53.5	70.0	61.1	63.2	63.5
		0 ∼ 2.5	57.1	53.5	69.0	61.5	63.9	62.6
		0 ∼ 4	56.0	54.0	70.1	61.2	63.9	62.6
3	mean - kurt	0 ∼ 1	54.3	54.2	72.8	60.0	63.5	61.5
		0 ∼2.5	56.0	52.3	71.4	61.3	63.2	61.7
		0 ∼ 4	53.9	53.9	70.5	59.7	60.7	60.0
4	mean - skew	0 ∼ 1	57.1	54.0	73.8	60.5	63.8	59.0
		0 ∼ 2.5	56.5	55.2	72.2	60.6	63.2	60.5
		0 ∼ 4	56.2	54.6	71.2	59.9	62.5	61.6
5	min - slope	0 ∼ 1	53.6	57.1	72.0	60.9	64.7	61.9
		0 ∼ 2.5	54.4	56.7	73.4	59.3	65.2	62.8
		0 ∼ 4	55.1	57.3	74.0	59.7	64.8	62.3
6	min - skew	0 ∼ 1	57.2	53.6	73.9	61.0	65.2	58.9
		0 ∼ 2.5	56.6	54.3	72.7	59.6	63.9	60.8
		0 ∼ 4	55.1	54.7	69.5	58.4	63.8	63.9
7	min - kurt	0 ∼ 1	53.8	54.5	72.8	59.2	64.0	61.1
		0 ∼ 2.5	55.2	51.5	71.4	60.6	63.9	61.5
		0 ∼ 4	53.2	52.0	70.3	57.7	62.3	61.9
8	slope - skew	0 ∼ 1	57.3	55.9	71.9	59.3	62.0	59.3
		0 ∼ 2.5	56.6	57.2	73.4	57.3	61.9	60.3
		0 ∼ 4	56.5	56.8	72.8	57.7	62.3	62.3
9	slope - kurt	0 ∼ 1	51.7	54.0	69.1	54.8	62.7	60.4
		0 ∼ 2.5	54.7	54.1	70.1	54.5	63.6	58.6
		0 ∼ 4	54.5	56.2	71.6	55.1	60.7	59.1
10	skew - kurt	0 ∼ 1	56.4	52.2	71.7	58.2	59.7	58.7
		0 ∼ 2.5	58.3	51.5	70.2	59.2	60.0	61.3
		0 ∼ 4	53.8	54.1	68.6	56.2	60.8	62.9