Also published in Language Log:
I am a cat?
x hide permanently

Do STT systems have "intriguing properties"?

Language Log

« previous post | In "Intriguing properties of neural networks" (2013), Christian Szegedy et al. point out that … deep neural networks learn input-output mappings that are fairly discontinuous to a significant extent. We can cause the network to misclassify an image by applying a certain imperceptible perturbation… For example: There has been quite a bit of discussion of the topic since then.