Back in June, Google released a series of rather trippy computer-generated images created by its image recognition neural network.
The images — dubbed “the dreams of machines” by a number of media outlets — were created by feeding images into the neural network and then having it recognize and accentuate certain features.
As the process is repeated more and more times, the image becomes increasingly strange. In the image below, for example, the network was fed a painting of a knight on horseback and directed to look for animals within it:
Neural networks were created to simulate how neurons behave in the human brain when given some sort of sensory input, like an image or sound.
But deep neural networks are capable of more than just finding and accentuating patterns within images. In fact, they can actually be taught to “paint” like some of history’s most famous artists.
A team of researchers from the University of Tubingen in Germany recently created a neural network capable of learning different painting styles and then applying them to new images.
Check out the short video below to learn more about the network and see it in action:
Details of the research were published in a paper entitled, “A Neural Algorithm of Artistic Style.” In the paper, the researchers write,
“Here we introduce an artificial system based on a Deep Neural Network that creates artistic images of high perceptual quality… The system uses neural representations to separate and recombine content and style of arbitrary images, providing a neural algorithm for the creation of artistic images.”
Image: Gatys, Ecker, Bethge
Besides being widely-used for image recognition tools, neural networks have also been used to predict the weather and detect breast cancer.
You can learn more about how they work here.