Fooling Machine Learning: Notes on Adversarial Attacks
中文摘要
本文探讨了机器学习中的对抗性攻击，说明了微小的物理改动（如在停止标志上贴胶带）如何欺骗AI模型。
English Summary
This article explores adversarial attacks, explaining how minor physical modifications, like taping a stop sign, can deceive machine learning models.
Original Excerpt
Picture a stop sign. Someone has stuck a few strips of black and white tape across it. Cheap tape, the kind you would walk past without… Continue reading on Medium »