I read about some similar work in a tweet by @jbhall56 recently https://twitter.com/jbhall56/status/1152189269399810048?s=20 … it’s very interesting to see how this will impact existing and new ML-based products 
Evasive attacks on ML classifiers are scary but we know little about adversarial ML in the problem space: we may need to rethink the whole formalization, incl. optimization and constraints for a principled approach - come and talk to us @USENIXSecurity poster session #usesec19pic.twitter.com/WxaWElHiWa
-
-
-
Thanks
@oerd, we were told about this today too - looking forward to reading that! - Još 1 odgovor
Novi razgovor -
-
-
Whoah. The poster looks amazing.
-
Feargus did all the characters from scratch with keynote shapes. It’s crazy awesome! :)
@0xtrustypatches - Još 1 odgovor
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.