A collection of 43 NetFlow features derived from the UNSW-NB15 network traffic traces, representing an expanded version of the original NF-UNSW-NB15 collection. It provides flow-level statistics and labels from the University of Queensland for identifying various types of network attacks and normal traffic patterns.
Use Cases
- Train binary classifiers to distinguish between normal and malicious traffic using the 43 NetFlow features
- Perform multi-class classification to identify specific attack categories based on flow-level statistics
- Evaluate feature importance across the 43 NetFlow metrics to optimize the performance of intrusion detection systems
- Benchmark machine learning models on standardized NetFlow data to ensure interoperability with real-world network monitoring tools
Strengths
- Includes 43 distinct NetFlow features such as flow duration, byte counts, and packet counts
- Serves as an expanded version (V2) of the NF-UNSW-NB15 dataset with a more granular feature set
- Derived from the original UNSW-NB15 network traffic traces for consistency in intrusion detection benchmarking
- Developed by researchers at the University of Queensland using standardized NetFlow extraction techniques