Table 7. Accuracy of classification on different selected sets of features.

Numbers of features	Accuracy (%)	Training time (s)
1,000	93.5	29.609
2,000	94.0	33.149
3,000	95.9	66.200
4,000	96.2	85.856
5,000	97.0	109.880
5,120 (50%)	97.6	111.690
2,560 (25%)	95.7	55.151
6,000	96.9	144.810
7,000	96.9	166.360
8,000	96.6	196.040