Skip to main content

Table 2 Evolving Attacks in the UNSW_NB15 dataset

From: Conceptualisation of Cyberattack prediction with deep learning

Attack Class

Description

Number of Instances

Train

Test

Normal

Benign traffic

56,000

37,000

Analysis

Can be a port scan or spam

2000

677

Backdoor

Bypassing a secured authentication to have access to a machine

1746

583

DoS

See Table 1

12,264

4089

Exploits

Exploitation of a vulnerability in a piece of software

33,393

11,132

Fuzzers

Feeding a network with randomly generated data to cause it to malfunction

18,184

6062

Generic

An attack on block ciphers

40,000

18,871

Reconnaissance

An attack that extracts information about a user or network

10,491

3496

Shellcode

Exploitation of a software vulnerability using a small piece of code as payload

1133

378

Worms

An attack that can replicate itself across multiple connected systems or networks

130

44