Table 3 Statistical analysis of the dataset used in our experiment.

From: A federated transformer-enhanced double Q-network for collaborative intrusion detection

Dataset

Label type and attack classes

Samples

NF-BoT-IoT

Binary: Normal

13,859

Binary: Attack

586,241

Multi: Benign

13,859

Multi: Theft

1909

Multi: DoS

56,833

Multi: DDoS

56,844

Multi: Reconnaissance

470,655

NF-ToN-IoT

Binary: Normal

270,279

Binary: Attack

1,108,995

Multi: Benign

270,279

Multi: Ransomware

142

Multi: MITM

1295

Multi: DoS

17,717

Multi: Backdoor

17,247

Multi: Scanning

21,467

Multi: XSS

99,944

Multi: Password

156,299

Multi: DDoS

326,345

Multi: Injection

468,539

NF-UNSW-NB15

Binary: Normal

2,295,222

Binary: Attack

95,053

Multi: Benign

2,295,222

Multi: Exploits

31,551

Multi: Fuzzers

22,310

Multi: Generic

16,560

Multi: Reconnaissance

12,779

Multi: DoS

5794

Multi: Analysis

2299

Multi: Backdoor

2169

Multi: Shellcode

1427

Multi: Worms

164

NF-CICIDS2018

Binary: Normal

16,635,567

Binary: Attack

2,258,141

Multi: Benign

16,635,567

Multi: DDOS-HOIC

1,080,858

Multi: DoS-Hulk

432,648

Multi: DoS-LOIC-HTTP

307,300

Multi: Bot

143,097

Multi: Infiltration

116,361

Multi: SSH-Brute

94,979

Multi: DoS-GoldenEye

27,723

Multi: FTP-Brute

25,933

Multi: DoS-SlowHTTP

14,116

Multi: DoS-Slowloris

9512

Multi: Brute-Web

2143

Multi: DDOS-LOIC-UDP

2112

Multi: Brute-XSS

927

Multi: SQL injection

432

Back to article page