Tuesday, 6. December 2016

You can train neural networks to generate images. This is done by combining two networks – one learns to recognize images, while the other learns to generate images that can fool it. The networks are adversaries and the technique is called Deep Convolutional Generative Adversarial Networks (DCGAN).

This is part of the project Principal Components where we are investigating the use of machine learning technologies with the Norwegian National Museum in collaboration with Audun Mathias Øygard. The project is supported by the Norwegian Arts Council.

Given that national romantic landscapes figure prominently in the collection of the National Museum it was tempting to train a network generate new samples.