Amazon's new algorithm designs clothing by analyzing a bunch of pictures

Amazon has developed an algorithm that can design clothing by analyzing a bunch of images to copy the style and then apply it to new items generated from scratch, reports MIT. Amazon is focussing heavily on fashion, and has previously said it's looking to become "the best place to buy fashion online," so designing its own pieces pinned to current trends seems like a logical move. The algorithm was developed by Amazon Lab 126, the company's research and development hub and uses a tool called generative adversarial network (GAN). The GAN is a new technique in AI research that forces two neural networks against one another, using the outcome to improve the overall system. It learns from raw data -- so by looking at a batch of images, it can draw out a style. That could mean spotting a particular trend on Instagram, Facebook, or its own collection of fashion imagery generated by the company's new Echo Look cameras.