Table 10.
Performance Analysis of four feature evaluation measures on UNSW-NB15.
| Attack class | ALL features |
Correlation |
Consistency |
Information |
Distance |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DR | FAR | KS | DR | FAR | KS | DR | FAR | KS | DR | FAR | KS | DR | FAR | KS | |
| Fuzzers | 0.937 | 0.145 | 0.780 | 0.799 | 0.267 | 0.659 | 0.941 | 0.143 | 0.784 | 0.909 | 0.196 | 0.710 | 0.860 | 0.176 | 0.653 |
| Analysis | 0.999 | 0.116 | 0.926 | 0.998 | 0.186 | 0.870 | 0.999 | 0.11 | 0.931 | 0.997 | 0.125 | 0.897 | 0.996 | 0.142 | 0.879 |
| Backdoor | 1 | 0.037 | 0.976 | 1.000 | 0.183 | 0.895 | 1 | 0.019 | 0.982 | 0.999 | 0.042 | 0.963 | 0.999 | 0.091 | 0.943 |
| DoS | 0.997 | 0.018 | 0.979 | 1.000 | 0.211 | 0.860 | 0.996 | 0.02 | 0.971 | 0.998 | 0.011 | 0.982 | 0.993 | 0.129 | 0.897 |
| Exploits | 0.991 | 0.012 | 0.978 | 0.979 | 0.282 | 0.801 | 0.990 | 0.015 | 0.975 | 0.982 | 0.034 | 0.947 | 0.972 | 0.092 | 0.925 |
| Generic | 0.999 | 0.004 | 0.996 | 0.999 | 0.083 | 0.983 | 0.999 | 0.004 | 0.995 | 0.999 | 0.007 | 0.990 | 0.998 | 0.027 | 0.987 |
| Reconnaissance | 0.999 | 0.006 | 0.992 | 0.988 | 0.126 | 0.928 | 0.997 | 0.014 | 0.981 | 0.995 | 0.029 | 0.965 | 0.995 | 0.084 | 0.967 |
| Shellcode | 0.998 | 0.125 | 0.893 | 0.986 | 0.224 | 0.619 | 0.999 | 0.090 | 0.916 | 0.995 | 0.425 | 0.618 | 0.995 | 0.473 | 0.582 |
| Worms | 1 | 0.208 | 0.861 | 1.000 | 0.292 | 0.753 | 1.000 | 0.062 | 0.981 | 1.000 | 0.069 | 0.920 | 1.000 | 0.123 | 0.904 |