Humans can decipher adversarial images

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 s...

Full description

Saved in:
Bibliographic Details
Published in:Nature communications 2019-03, Vol.10 (1), p.1334-1334, Article 1334
Main Authors: Zhou, Zhenglong, Firestone, Chaz
Format: Article
Language:English
Subjects:
631/378/3917
631/477/2811
639/705/1042
639/705/117
Artificial neural networks
Benchmarks
Choice Behavior
Classification
Eye (anatomy)
Foils
Human behavior
Humanities and Social Sciences
Humans
Image classification
Image Processing, Computer-Assisted
Imaging, Three-Dimensional
Judgments
Learning algorithms
Machine learning
multidisciplinary
Neural networks
Object recognition
Pattern recognition
Science
Science (multidisciplinary)
Television
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary: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. Convolutional Neural Networks (CNNs) have reached human-level benchmarks in classifying images, but they can be “fooled” by adversarial examples that elicit bizarre misclassifications from machines. Here, the authors show how humans can anticipate which objects CNNs will see in adversarial images.
ISSN:2041-1723
2041-1723
DOI:10.1038/s41467-019-08931-6