Apple publishes its first paper on artificial intelligence: Report

The paper describes a technique that improves the training that can given to an algorithm to recognise images using computer-generated images as opposed to real world. According to the research, in machine learning, using computer generated images in training neutral networks is more efficient than real-world images, primarily because these picture are already labelled authenticated. “Synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions”, the paper says.

The paper proposes that a solution to that problem what Apple researchers call “Simulated+Unsupervised (S+U) learning”.  The Apple researchers use a custom version of Generative Adversarial Networks (GANs), through which  the realism of a simulated image is boosted.

Also read: Apple’s machine learning system, and why it won’t sacrifice user privacy: Report

Artificial intelligence (AI) is said to be the next big thing in technology. Microsoft, Google, IBM, Amazon and Facebook recently announced the creation of “Partnership on AI”. Called “Partnership on Artificial Intelligence to Benefit People and Society”, it will be primarily aimed at educating users about artificial technologies.

While its peers in the tech industry Google and Microsoft have been betting on artificial intelligence, Apple hasn’t been vocal about the new emerging technology. But looks like Apple is working to improve the machine-leaning techniques for quite sometime.

Apple has big plans to integrate all its devices through the Siri, which can be found on its iOS devices, MacOS, Apple TV and watchOS. In September, Bloomberg’s Mark Gurman revealed that Apple could be working on Siri-powered Amazon Echo speaker. Bloomberg reported that the intelligent speaker will work with Apple Home and Homekit-powered devices.