Originally posted by Sysman
View Post
The key to Falcon is that it generates an understanding from n-thousand dimensions of what is "normal" for a particular card, not for the whole portfolio. If it was done on a portfolio basis you'd be getting a much higher false-positive rate - here, the false positive rate with Falcon is around 8:1 - in the non-ANN environment, it's around 55:1 at best.
Comment