Deep Flaw – 2016
World’s first AI generated and evolutionary driven art (sculptures_3d prints).
Combining EC and AI’s to generate 3d models. The custom system is intended to be used with an input of minimum 8 “snapshot” images of an object, to start evolving. Here we see evolved source images of a piggybank and Mickey Mouse rendered in a 3d model that an AI still recognises as a piggybank and Mickey Mouse but look utterly strange to us.
Advances in supervised learning with deep neural net- works have enabled robust classification in many real world domains. An interesting question is if such advances can also be leveraged effectively for computational creativity. One insight is that because evolutionary algorithms are free from strict requirements of mathematical smoothness, they can exploit powerful deep learning representations through arbitrary computational pipelines. In this way, deep networks trained on typical supervised tasks can be used as an ingredient in an evolutionary algorithm driven towards creativity.
The conclusion is that combining EC and deep learning in this way provides new possibilities for creative generation of meaningful and novel content from large labeled datasets.
Credits
Deep Flaw is a project by Frederik De Wilde in an ongoing collaboration with Joel Lehman, Jeff Clune and Alexander Holdt. De Wilde already collaborated with Jeff Clune and Anh Nguyen on “The Innovation Engine.”
Partnering Universities
Center for Computer Games Research IT University of Copenhagen and the Department of Computer Science University of Wyoming Laramie, Wyoming, USA.)
Ref. Creative Generation of 3D Objects with Deep Learning and Innovation Engines