All data were randomly selected from the CSE-CIC-IDS2018 dataset. The data fields were censored after going through the analysis and 64 valid features were retained.
There are 5 types of data, they are Benign, DoS, DDoS, Botnet and Infiltration.
There are five types of data in the dataset, namely NORMAL, DoS, Probe, R2L and U2R. A total of 20,000 training samples were used during the experiment (5 classifications in total, 4000 samples for each classification). There are 4047 samples in the validation dataset, including 1000 samples each of NORMAL, DoS, and Probe types. 995 samples of R2L and 52 samples of U2R.