Dirty Data (training with dirty data)

Brief description

The artist is training a neural network to generate images. However, instead of cleaning his dataset as it is a practice when generating hyper-realistic images of persons, he leaves all images tagged with a politician’s name even if it does not represent the person in questions face. As a result, the machine vision assemblage generates is creepy satires of something that reminds us of a face. In the video work presents procedural iterations of machine learning and generated images.

Work that the situation appears in

Title Publication Type Year Creator
Dirty Data Art, Video art Memo Akten
Who does what?
Aesthetic characteristics
Machine P.O.V
Not machine P.O.V.

Authored by

UUID
145aec91-5491-4a81-8c06-1a5b75c32e22