Nicole Junkermann presents an A-Z of AI: G = GANs

Nicole Junkermann, NJF Capital founder, presents the next video in the series ‘an A-Z of Artificial Intelligence’ – a selection of short videos focused on key areas of interest in this hugely topical and consequential field.

Generative Adversarial Networks (GANs) are a type of AI that uses a pair of neural networks to generate content or perform other tasks.

For example, GANs have been trained to create realistic images of human faces after analysing banks of celebrity portraits.

One neural network, the generator, generates images, while another, the discriminator, analyses them for authenticity and suggests improvements.

By repeating this process millions of times, the generator learns to create convincing images.

Applications for GANs include image enhancement, 3D modelling and computer game design.

Scientists have used GANs to enhance astronomical images from deep space.

French art collective Obvious used a GAN to create the picture Edmond De Belamy, which was sold for $432 500 at Christies New York in 2018.