Department of Psychological & Brain Sciences, Johns Hopkins University, 3400 N Charles St., Baltimore, MD, 21218, USA.
Nat Commun. 2019 Mar 22;10(1):1334. doi: 10.1038/s41467-019-08931-6.
Does the human mind resemble the machine-learning systems that mirror its performance? Convolutional neural networks (CNNs) have achieved human-level benchmarks in classifying novel images. These advances support technologies such as autonomous vehicles and machine diagnosis; but beyond this, they serve as candidate models for human vision itself. However, unlike humans, CNNs are "fooled" by adversarial examples-nonsense patterns that machines recognize as familiar objects, or seemingly irrelevant image perturbations that nevertheless alter the machine's classification. Such bizarre behaviors challenge the promise of these new advances; but do human and machine judgments fundamentally diverge? Here, we show that human and machine classification of adversarial images are robustly related: In 8 experiments on 5 prominent and diverse adversarial imagesets, human subjects correctly anticipated the machine's preferred label over relevant foils-even for images described as "totally unrecognizable to human eyes". Human intuition may be a surprisingly reliable guide to machine (mis)classification-with consequences for minds and machines alike.
人类的思维是否类似于反映其性能的机器学习系统？卷积神经网络 (CNN) 在对新型图像进行分类方面已达到人类水平的基准。这些进展支持了自动驾驶汽车和机器诊断等技术的发展；但除此之外，它们本身也可以作为人类视觉的候选模型。然而，与人类不同的是，CNN 会被对抗性示例所“愚弄”——这些示例是机器识别为熟悉物体的无意义模式，或者是看似不相关的图像干扰，但却改变了机器的分类。这些奇怪的行为挑战了这些新进展的承诺；但是人类和机器的判断是否从根本上不同呢？在这里，我们表明，人类和机器对对抗性图像的分类是紧密相关的：在对 5 个著名的、多样化的对抗性图像集进行的 8 项实验中，人类受试者能够正确地预测出机器对相关干扰的首选标签，即使是对于被描述为“人类的眼睛完全无法识别”的图像也是如此。人类的直觉可能是机器（错误）分类的一个惊人可靠的指南——这对人类和机器都有影响。
