Research Article
GTF: An Adaptive Network Anomaly Detection Method at the Network Edge
Table 3
Class distribution and
of UNSW-NB15.
| | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Num. | Num. | | Num. | | Num. | | Num. | | Num. | | Num. | | Num. | | Num. | | Num. | |
| Training | 56 000 | 10 491 | 5.3 | 1746 | 32.1 | 12 264 | 4.6 | 33 393 | 1.7 | 2000 | 28.0 | 18 184 | 3.1 | 130 | 430.8 | 1133 | 49.4 | 40 000 | 1.4 | Testing | 37 000 | 3496 | 10.6 | 583 | 63.5 | 4089 | 9.0 | 11 132 | 3.3 | 677 | 54.7 | 6062 | 6.1 | 44 | 840.9 | 378 | 97.9 | 18 871 | 2.0 |
|
|
0: Normal, 1: Reconnaissance, 2: Backdoor, 3: DoS, 4: Exploits, 5: Analysis, 6: Fuzzers, 7: Worms, 8: Shellcode, and 9: Generic. |