4,898,431 connection records categorized into 23 attack types and a 'normal' class. The data includes 41 features per connection, such as protocol_type, service, and src_bytes, derived from raw TCP dump data.
Use Cases
- Train a multi-class classifier to predict the 'label' column using 41 network features
- Analyze the correlation between 'src_bytes' and 'dst_bytes' to detect Denial of Service (DoS) attacks
- Evaluate anomaly detection algorithms by comparing 'normal' traffic against the four attack categories
Strengths
- 4,898,431 connection records representing seven weeks of network traffic
- 41 features per record, including basic connection attributes like duration and protocol_type
- Labels for 23 attack types categorized into DoS, R2L, U2R, and Probing