Why the NSL-KDD

■ A dataset of network traffic records 

■ Created in 2009, curated from KDDCup99 (1999)

■ Pros:

– One of the few publicly available datasets
– Already labelled (necessary for supervised learning)
– Rich in features
– Still used in research 

■ Cons:

– Outdated (doesn’t contain recent attacks)
– Synthetic dataset

9