The Nightmare Machine

This is not from the Nightmare Machine, but from the 20th century horror-artist Francis Bacon. Surely he could teach Google's AI a thing or two.

AI creates gallery of nightmare images for Halloween
Oct 2016, BBC

The creators of the Nightmare Machine are trying to "help our algorithms learn scariness." In other words, they're teaching machines how to scare us. They called it the Nightmare Machine, not me. And if this isn't enough to scare Stephen Hawking on Halloween, I don't know what is.

Two techniques they're using are called style transfer and generative adversarial networks; just mentioning this because they sound cool.

POST SCRIPT

On Google AI's previous project, Deep Dream:

Inceptionism vs Trypophobia
Network Address, 2015

Deep Bosch
Network Address, 2015

Inceptionism vs Trypophobia

Left my computer dreaming overnight, found this when I woke up

gallery

Deep Learning in Reverse Shows You What an Algorithm Sees
#DeepDream

Phantasmagoric neural net visions
mindhacks.com, Jun 2015

All I have to say is, sucks to be trypophobic right now.

Deep Bosch

Hieronymus Bosch’s “The Garden of Earthly Delights”, Kyle McDonald/Flickr

Google’s New Visualization Tool Slips Slimy and Furry Creatures into Art History
Claire Voon, Hypoallergic, July 7, 2015

Inceptionism
Each layer of the network deals with features at a different level of abstraction, so the complexity of features we generate depends on which layer we choose to enhance. For example, lower layers tend to produce strokes or simple ornament-like patterns, because those layers are sensitive to basic features such as edges and their orientations.

"...overinterpret [...] oversaturated with snippets of other images."

Hieronymus Bosch’s “The Garden of Earthly Delights”, @aut0mata /Twitter

Phantasmagoric neural net visions
mindhacks.com, Jun 2015

[mindhacks always has the best explanations]

by using the neural networks “in reverse” they could elicit visualisations of the representations that the networks had developed over training.

pictures are freaky because they look sort of like the things the network had been trained to classify, but without the coherence of real-world scenes

The obvious parallel is to images from dreams or other altered states – situations where ‘low level’ constraints in our vision are obviously still operating, but the high-level constraints – the kind of thing that tries to impose an abstract and unitary coherence on what we see – is loosened. In these situations we get to observe something that reflects our own processes as much as what is out there in the world.

link
gallery

***

the code has been opened for all to use; let the dreaming begin
#DeepDream