Apple Launches Machine Learning Journal To Boost AI Research

Apple’s Director of AI Research, Russ Salakhutdinov appears to have persuaded Apple to be a little more open concerning its plans for Machine Learning, convincing the company to launch a new online publication to deal with the company’s work in developing AI, the ‘Machine Learning Journal’.

The first post on the (currently very small) site states:

“Welcome to the Apple Machine Learning Journal. Here, you can read posts written by Apple engineers about their work using machine learning technologies to help build innovative products for millions of people around the world.”

Apple’s Machine Intelligence team quite plainly wants to communicate with peers to develop these next-generation technologies. “If you’re a machine learning researcher or student, an engineer or developer, we’d love to hear your questions and feedback,” the site declares.

The first full publication on the Journal (which I guess Apple’s team hopes will become a valuable peer within the AI research ecosystem) discusses a way to train machine intelligence using synthetic images. It has been thought that using such synthetic images to train machines is inadequate to creating accuracy, however  Apple’s teams have got past this problem.

“We show that training models on these refined images leads to significant improvements in accuracy on various machine learning tasks,” the report said.

Apple recruited Carnegie Mellon University Associate Professor Salakhutdinov last year. He said he was “excited” in his new role.

Machine intelligence is a work in progress at every institution and company so it’s no great surprise that in order to put development on the fast track Apple now appears convinced to allow its highly academic research team to work in the traditional peer-reviewed and collaborative way. Most people understand that innovation is international and requires some sharing of resources.

Apple presented and published its first academic paper on artificial intelligence at an industry conference last year. This latest report appears to be its second such paper. What’s interesting about this is that it seems to show a way in which machine intelligence can gather useful insights without actually demanding use of your private data.

Image c/o Mini Ozzy/Flickr

Jonny Evans

Watching Apple since 1999. I don't say what they should do. I say what they might do. They sometimes do.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.