Apple publishes research paper about artificial intelligence

Apple publishes research paper about artificial intelligence

Last week, tech giant Apple published its first research paper related to artificial intelligence (AI). The paper was published by a team of Apple researchers through the Cornell University Library.

The first AI research paper published by Apple is titled ‘Learning from Simulated and Unsupervised Images through Adversarial Training.’ The paper – which describes a technique for improving AI – was submitted for review on November 15, 2016. The paper has been authored by Apple researchers Ashish Shrivastava, Tomas Pfister, Josh Susskind, Oncel Tuzel, Russ Webb, and Wenda Wang.

The key focus of Apple’s AI research paper is on bringing about an improvement in a computer's ability to recognize images --- a task which was extremely difficult for computers in the past.

The paper proposes the use of what it called "Simulated+Unsupervised (S+U)" learning for improving the capability of a computer to recognize images.

The core component of the paper is the use of digital images which have data within them, so that computers are able to "see" images instead of real-world objects that require interpretation. In addition, for improving a computer’s image-recognition capabilities, the paper also suggests the modification of certain AI algorithms.