Our proposed work shows NIPS attendees executing various high level manipulations of their own face images using generative neural networks. Portrait photos are encoded into the latent space of a variational autoencoder where attribute vectors can be applied. These include opening and closing the mouth, or adding or removing a smile. Images are then decoded from the latent space and videos are created showing these effects. Additionally, participants can define their own attribute vector by having two photos taken and using the difference between them. This new attribute vector can then be applied to provided reference images as a one-shot generalization.
Tom White (Victoria University of Wellington School of Design)
Tom is a New Zealand based artist investigating machine perception. His current work focuses on creating physical artworks that highlight how machines “see” and thus how they think, suggesting that these systems are capable of abstraction and conceptual thinking. He has exhibited computer based artwork internationally over the past 25 years with themes of artificial intelligence, interactivity, and computational creativity. He is currently a lecturer and researcher at University of Wellington School of Design where he teaches students the creative potential of computer programming and artificial intelligence.
More from the Same Authors
2020 Workshop: Machine Learning for Creativity and Design 4.0 »
Luba Elliott · Sander Dieleman · Adam Roberts · Tom White · Daphne Ippolito · Holly Grimm · Mattie Tesfaldet · Samaneh Azadi
2019 Workshop: NeurIPS Workshop on Machine Learning for Creativity and Design 3.0 »
Luba Elliott · Sander Dieleman · Adam Roberts · Jesse Engel · Tom White · Rebecca Fiebrink · Parag Mital · Christine Payne · Nao Tokui
2018 Workshop: Second Workshop on Machine Learning for Creativity and Design »
Luba Elliott · Sander Dieleman · Rebecca Fiebrink · Jesse Engel · Adam Roberts · Tom White