Apple Lossless Audio CODEC (ALAC), is one of many American companies developing artificial intelligence systems to improve the way its offline and online services work. Recently, Apple has allowed the researchers working on this system to publish their own scientific works to explain methods by which artificial intelligence can evolve.
Well, now even Apple has published a scientific work of this kind, the American company presenting a small part of what it is preparing for the future. In its scientific work, Apple talks about the method by which artificial intelligence can learn by analyzing the images that are offered to it by the creators.
What Apple describes is most likely part of the system implemented in iOS 10 to analyze the images recorded with the iPhone, iPad and iPod Touch and sort them. Apple describes the benefits and downsides of using real images compared to digitally created ones, with artificial intelligence needing annotations to understand objects in real images.
Those from Apple try to present in as few details as possible what they are developing at the moment in the field of artificial intelligence, in order not to offer their competitors help in developing their own technologies. If you are interested in reading the scientific paper published by the Apple company, well you can find it in this page.
"In this paper, we propose Simulated+Unsupervised (S+U) learning, where the goal is to improve the realism of synthetic images from a simulator using unlabeled real data. The improved realism enables the training of better machine learning models on large datasets without any data collection or human annotation effort. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study."
When will the contest winners be announced?
Tomorrow, most likely.