/* ---- Google Analytics Code Below */

Friday, September 04, 2020

Machine Generated Art

Automated art has been around for a long time, but it has never clicked in a substantial way.  Will some of these advances change that?  Is the case where such attempts should really be more direct collaboration with humans? 

Rewriting the Rules of Machine-Generated Art

MIT News
Kim Martineau
August 18, 2020

Massachusetts Institute of Technology (MIT) researchers have shown it is possible to edit deep layers of neural networks to generate images never seen before. Generative adversarial networks (GANs) typically are trained on massive datasets, but MIT’s study suggests large datasets are not essential to the process. Said MIT's David Bau, "We’re like prisoners to our training data. GANs only learn patterns that are already in our data, but here I can manipulate a condition in the model to create horses with hats. It’s like editing a genetic sequence to create something entirely new, like inserting the DNA of a firefly into a plant to make it glow in the dark.” The tool has immediate applications in computer graphics, and in teaching expert AI systems to recognize rare features and events through data augmentation. ... " 


No comments: