Navigation
Recherche
|
Can your AI code be fooled by vandalized images or clever wording? Microsoft open sources a tool to test for that
jeudi 6 mai 2021, 01:27 , par TheRegister
Counterfit automatically creates adversarial inputs to find weaknesses
Microsoft this week released a Python tool that probes AI models to see if they can be hoodwinked by malicious input data.…
https://go.theregister.com/feed/www.theregister.com/2021/05/05/microsoft_ai_security/
|
56 sources (32 en français)
Date Actuelle
sam. 4 mai - 17:48 CEST
|