Table 1:
Comparison of Test Error (%) of Models 
and 
on UCI Data Sets.
and on UCI Data Sets. |  |  |  | |
|---|---|---|---|---|
 | 19.53 [20.02(1.53)] | 26.09 [30.96(3.34)] | 0 [0.68(1.60)] | 16.99 [17.46(0.65)] |
 | 18.75 [19.39(0.89)] | 25.22 [31.48(4.11)] | 0 [5.38(12.0)] | 17.08 [17.48(0.56)] |
 | 21.88 [24.73(5.91)] | 28.69 [32.39(3.48)] | 0 [8.21(6.07)] | 17.83 [21.21(4.78)] |
 | 21.09 [20.20(1.51)] | 27.83 [28.86(2.83)] | 1.71 [5.68(2.22)] | 15.07 [15.15(0.15)] |
 | 18.75 [20.33(2.75)] | 28.69 [28.38(2.80)] | 10.23 [6.92(3.69)] | 14.91 [15.08 (0.15)] |
 | 19.03 [19.16(1.10)] | 26.08 [27.74(9.40)] | 6.83 [6.50(8.31)] | 15.71 [15.97(0.07)] |
 | 24.61 [22.34(1.95)] | 32.17 [27.61(3.69)] | 3.42 [9.66(6.74)] | 15.21 [15.19(0.08)] |
| bestbmark | 22.7(2.2) | 29.6(3.7) | 4.0(2.1) | 14.4(0.3) |
| | | | |
|---|---|---|---|---|
| | 19.53 [20.02(1.53)] | 26.09 [30.96(3.34)] | 0 [0.68(1.60)] | 16.99 [17.46(0.65)] |
| | 18.75 [19.39(0.89)] | 25.22 [31.48(4.11)] | 0 [5.38(12.0)] | 17.08 [17.48(0.56)] |
| | 21.88 [24.73(5.91)] | 28.69 [32.39(3.48)] | 0 [8.21(6.07)] | 17.83 [21.21(4.78)] |
| | 21.09 [20.20(1.51)] | 27.83 [28.86(2.83)] | 1.71 [5.68(2.22)] | 15.07 [15.15(0.15)] |
| | 18.75 [20.33(2.75)] | 28.69 [28.38(2.80)] | 10.23 [6.92(3.69)] | 14.91 [15.08 (0.15)] |
| | 19.03 [19.16(1.10)] | 26.08 [27.74(9.40)] | 6.83 [6.50(8.31)] | 15.71 [15.97(0.07)] |
| | 24.61 [22.34(1.95)] | 32.17 [27.61(3.69)] | 3.42 [9.66(6.74)] | 15.21 [15.19(0.08)] |
| bestbmark | 22.7(2.2) | 29.6(3.7) | 4.0(2.1) | 14.4(0.3) |
Notes: Shown first is the test error corresponding to the selected model with minimal validation error from the different random initializations. Between brackets, the mean and standard deviation of the test errors related to all initializations are shown. The lowest test error is in bold.