search

FabienCharmet / XAISurvey

0 stars 0 forks source link

marino2018adversarial: Criticizing the dataset #2

Open FabienCharmet opened 2 years ago

FabienCharmet commented 2 years ago

I reviewed marino2018adversarial and I criticized the NSL-KDD dataset.

I don't know if it's relevant or not.

There a some papers who use it so...

cloudgravity commented 2 years ago

NSL-KDD is a replacement for KDD99, removing many biases. It is still quite old and not representative of recent attacks, and should not be used if authors want to tackle modern systems. It has nonetheless unique features such as the ones related to HIDS. If the paper relies on HIDS features, then it makes sense to use NSL-KDD in the absence of a more recent (yet relevant) dataset.