A fun piece of adversarial ML: the adversaries are *human* essay-writers trying to fool an NLP-powered essay rating system into giving high scores to crap essays... *or* low scores to subtly brilliant ones!
(sadly, I don't see the raw essays, but the descriptions still shine)
Quote Tweet
I was feeling pretty dispirited in the midst of a literature review on automatic essay scoring systems (
), but this paper on an
adversarial essay writing contest
fooling the scoring system totally made my day. doi.org/10.1002/j.2333
1
4
25

