Deep Image



—Deep Generator Network, generative adversarial network, sensors
—5.1 sound
–2021–ongoing

[Installation shot at Nogueras Blanchard, Barcelona 2021]

Produced by LEONARDO grant

LINK
LINK
DEEP IMAGE consists of an immersive video installation that, through collaboration with a programmer and a neuroscientist, will decode the activity of the human brain in sequences of images. Through an artificial intelligence 'software' developed in collaboration with the Kamitani Lab of the University of Kyoto and the Laboratory of Cognitive and Computational Neuroscience of the Polytechnic University of Madrid, an immense and complete conceptual map of brain images obtained from the memories, dreams or sounds of a group of people. 

Artificial intelligence will use convolutional neural networks -better known as 'deep learning' networks- and will be made up of internal layers of 'software' that will share information with each other to produce synthetic forms from the data obtained. 

The main idea of Deep Image is to reconstruct the images, thoughts and memories that are formed in the brain by a deep neural network, which visualizes the information from magnetoencephalography analysis (MEG) in a video installation. 

In order to do this, I collaborated with one of the leading laboratories in Europe, the Laboratory of Cognitive and Computational Neuroscience of the UCM University from Madrid and the the Kamitani Lab of the University of Kyoto, using magnetoencephalography techniques (MEG) to obtain all this information from 50 people.