You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello, I was trying ot reproduce the baseline presented in the paper "Deep Learning for Unsupervised Insider Threat Detection in Structured Cybersecurity Data Streams"
However the dataset containing count features provided in this repository seems to be corrupted, some rows are mixed, and some columns are shifted
could you point to the script used to produce the data, or to a non-corrupted dataset ?
Thanks
The text was updated successfully, but these errors were encountered:
Hello, I was trying ot reproduce the baseline presented in the paper "Deep Learning for Unsupervised Insider Threat Detection in Structured Cybersecurity Data Streams"
However the dataset containing count features provided in this repository seems to be corrupted, some rows are mixed, and some columns are shifted
could you point to the script used to produce the data, or to a non-corrupted dataset ?
Thanks
The text was updated successfully, but these errors were encountered: