Finally :')
We're releasing a new method to test for model robustness against adversaries not seen during training, and open-sourcing a new metric, UAR (Unforeseen Attack Robustness), which measures how robust a model is to an unanticipated attack: https://openai.com/blog/testing-robustness/ …pic.twitter.com/8yJdd6oD5T
-
-
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
I just saw Donal trump
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Really interesting take at evaluating adversarial robustness against unseen attacks! It would be interesting to see how this metric fares on a model trained by AMC (http://precog.iiitd.edu.in/pubs/Hardening_Deep_Neural_Networks_IJCNN.pdf …)
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
I like OpenAI, but it makes no sense for a model to recognize an input that was *unanticipated*! A model is either designed to handle an input, or not. The real problem to address is, "how do we make our computed models behave more like the human visual system?"
- Još 2 druga odgovora
Novi razgovor -
-
-
Very important step to go beyond generalization-only view of ML performance.
- Još 1 odgovor
Novi razgovor -
-
-
ML is of course in progress as it gets hard to test, but isn't there some people out there work'n on another approach than ML!!! I hope Research scientists have this very task
- 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.