Sign In

Communications of the ACM

ACM TechNews

How AI Is Creating Building Blocks to Reshape Music and Art

View as: Print Mobile App Share:
DeepDream artwork

An artwork created using DeepDream, which researchers at Google developed in 2015.

Interest in using deep neural networks to train machines to produce artwork by analyzing massive datasets is growing, and innovators such as Google researcher Douglas Eck envision artificial intelligence as a tool that artists can use to explore new modes of creation. For example, Eck's team is feeding musical notes to a neural network called NSynth so sounds from different instruments can be combined into new forms. In 2015, another Google team created DeepDream, which trained neural networks on photos as a basis for them to create haunting imagery.

Meanwhile, David Ha, who works with Eck's team on Google's Project Magenta, constructed a neural network that learned to compose original logographic Chinese characters. Ha currently is building neural networks with the ability to draw, by analyzing thousands of human-made sketches to mathematically identify what specific drawings look like, and then executing their own drawings in accordance with an artist's instructions.

From The New York Times
View Full Article – May Require Registration


Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account