Google DeepMind's Latest Research: Adversarial Attacks Are Effective Against Humans Too, Both Humans and AI Mistake Vases for Cats
新智元
39
Translated data:
Google DeepMind's latest research indicates that adversarial attacks are not only effective against artificial intelligence but can also influence human judgment. Neural networks are susceptible to adversarial perturbations, leading both humans and AI to misclassify images. This finding underscores the need for a deeper understanding of the behavioral differences and similarities between AI visual systems and human perception, in order to develop safer AI systems.
© Copyright AIbase Base 2024, Click to View Source - https://www.aibase.com/news/4743