“GANs are the secret weapon for AI companies and projects,” TechRepublic contributor Jay Garmon said.

TechRepublic’s Dan Patterson spoke to Garmon to discuss how companies must use generative adversarial networks (GANs) if they want to succeed using artificial intelligence (AI).

GANs are two different AI software packages that fight against each other in order improve one another. These networks became popular in 2014, primarily for use in video and image processing.

SEE: IT leader’s guide to the future of artificial intelligence (Tech Pro Research)

“You tend to have two different AI systems in this pit fight: One is the generator and one is the discriminator,” Garmon said. The discriminator’s job is to take a set of videos or images, read them, then make fake imitation copies. For example, if you have 50 images of stop signs, you can take the machine learning training algorithm and turn the images into 100 pictures of something similar to stop signs. Then, the discriminator’s job is determine which images are real stop signs, and which images are fake.

“The better the generator gets at fooling the discriminator, you can take that data and retrain the discriminator to get better at spotting fakes,” he said. “Back and forth it goes until both are really, really good at their jobs, without you having to go out and get 10,000 pictures of stop signs.”

The most practical application of GANs can be seen in astronomy. Researchers use this network to improve their radio astronomy images, or videos, since they only have a limited number of images and videos to choose from.

“Any startup out there, any company that wants to do image and video work, that isn’t sitting on literally millions or billions of hours or samples of images, you basically have to use GANs. You don’t have an option because the data is just too expensive to produce.”

Also see