photos – Artificial Intelligence https://ars.electronica.art/ai/en Ars Electronica Festival 2017 Tue, 28 Jun 2022 13:43:24 +0000 en-US hourly 1 https://wordpress.org/?v=4.9.6 Portraits of Imaginary People https://ars.electronica.art/ai/en/portraits-of-imaginary-people/ Tue, 08 Aug 2017 20:31:55 +0000 https://ars.electronica.art/ai/?p=3412

Mike Tyka (DE)

Portraits of Imaginary People explores the latent space of human faces by training an artificial neural network to imagine and generate portraits of non-existent people.

To do so, thousands of photos of faces from Flickr were fed to a type of neural network technique called a “generative adversarial network” (GAN). GANs work by using two neural networks playing an adversarial game: one (the “generator”) tries to generate increasingly convincing output, while a the second (the “critic”) tries to learn to distinguish real photos from generated ones.

At first both networks are poor at their respective tasks. But as the discriminator network starts to learn to predict fake from real, it keeps the generator on its toes, pushing it to generate harder and harder examples. As the generator gets better the discriminator also has to improve in turn, in order to keep up. With time, the generated output becomes increasingly realistic, as both adversaries try to outwit each other.

Credits

www.miketyka.com

]]> Machine Learning Porn https://ars.electronica.art/ai/en/machine-learning-porn/ Tue, 08 Aug 2017 07:11:37 +0000 https://ars.electronica.art/ai/?p=3590

Jake Elwes (UK)

Artificial intelligence and machine learning are fast becoming part of everyday life. Based on AI models currently used, among other things, in content moderation and surveillance, the artworks explore the “latent space” of the AI as it processes and imagines the world for itself, dreaming in the areas between and beyond what it has learnt from us.

In Machine Learning Porn a neural network has been trained using an explicit content model for finding pornography in search engines. The network is then reverse engineered to generate new “pornography” from scratch: an AI daydreaming of sex.

Credits

www.jakeelwes.com
Special thanks to Gabriel Goh for inspiration.

]]> hananona https://ars.electronica.art/ai/en/hananona/ Tue, 08 Aug 2017 05:47:11 +0000 https://ars.electronica.art/ai/?p=1787

STAIR Lab. (JP) collaborating with Surface & Architecture Inc, Kyoko Kunoh, Tomohiro Akagawa, Tanoshim Inc., mokha Inc. and Tokyo Studio Co. Ltd. (JP)

The latest AI research makes it possible to teach computers the names of things by showing them many examples. The key is a large amount of training data and deep learning software. By leveraging this, the artists have developed an AI capable of classifying 406 kinds of flower by using over 300,000 flower pictures.

hananona is an interactive work that visualizes how AI classifies a flower. When it sees a flower, it identifies its name and shows its class on a visual “flower map”—a visualization of the inside of the AI brain. This is a group of image clusters, each of which is a cluster of flower photos learned as belonging to the same class. By looking at them, users can see how AI classifies the flowers.

Users are encouraged to challenge hananona with their own flower photos, or with other materials such as pictures, paintings, flower-like objects etc. so that they can observe how the AI reacts to different abstraction levels of flowers.

Credits

STAIR Lab., Chiba Institute of Technology

Creative direction, design: Surface & Architecture Inc.

Art direction: Kyoko Kunoh
Interaction design, programming: Tomohiro Akagawa
Programming: Tanoshim Inc.
Server programming: mokha Inc.
Furniture production, site setup: Tokyo Studio Co., Ltd.

]]>