Apple once said it would start publishing its own articles on Artificial Intelligence, without the announcement having completed a month, the company sent out its article related to the subject. The first article describes a technique that aims to improve the skill training of an algorithm that allows you to recognize images using computer generated images instead of real-world images.
According to the Apple article, in machine learning research, using synthetic images (such as those in a video game) to train neural networks can be more efficient than using real-world images. This is because synthetic image data is already labeled and annotated, while real-world image data requires someone to thoroughly check everything the computer is seeing.
The article also says that the approximation of the synthetic image can be a problem, because what the algorithm learns does not always load in order to the scenes of the real world. Synthetic image data "is often not realistic enough, leading the network to learn details only present in synthetic images and fail to generalize well in real images".
To solve the problem, the article suggests what Apple researchers call Simulation +, an unsupervised learning, where the realism of a simulated image is driven. Apple researchers use a modified version of a new machine learning technique called Generative Adversarial Networks, which places two neural networks together and has been used to generate photo-realistic images.
After this first work on AI, Apple marks a big step, since for years the Artificial Intelligence research community has criticized Apple's secrecy. His secret further hampered the company's recruiting efforts for AI talent.
What did you think of Apple's initiative?
