Table 2.

Recognition accuracy by n-fold cross validation procedure for different feature extraction techniques for SVM classification for the TG dataset.

Feature sets	n = 5	n = 6	n = 7	n = 8	n = 9	n = 10
PF1 [23]	38.1	38.4	38.6	38.7	38.8	38.8
PF2 [23]	38.0	38.4	38.5	38.6	38.7	38.8
PF [24]	42.3	42.6	42.7	43.0	43.0	43.1
O [21]	35.8	36.1	36.2	36.1	36.3	36.3
AAC [5]	31.5	31.5	31.7	31.8	31.9	32.0
AAC+HXPZV [5]	35.7	36.0	36.1	36.2	36.3	36.3
ACC [29]	64.9	65.4	65.9	66.2	66.4	66.4
PSSM+PF1 [55]	51.1	51.5	52.0	52.3	52.4	52.7
PSSM+PF2 [55]	50.2	50.4	50.7	50.8	51.0	51.1
PSSM+PF [55]	57.2	57.8	58.0	58.3	58.5	58.8
PSSM+O [55]	46.0	46.3	46.5	46.5	46.7	46.7
PSSM+AAC [55]	43.2	43.5	43.6	43.8	43.8	44.0
PSSM+AAC+HXPZV [55]	45.6	45.9	46.0	46.2	46.3	46.6
Mono-gram [19]	57.2	57.3	58.2	58.4	58.8	58.8
Bi-gram [19]	67.1	67.5	67.6	67.8	68.1	68.1
k -AAP (this paper)	75.9	76.2	76.6	76.7	76.9	77.0