Apple launches online journal for engineers to share their machine learning research

Apple has launched the ‘Machine Learning Journal’, a blog for Apple’s software engineers to document their research and innovations in the AI and machine learning space. The first post shares insights into a published research paper about improving picture-based machine learning models without gathering additional training data manually.

Apple will update the site throughout the year with new posts and is inviting questions from outside engineers, students and researchers.

The first post is lengthy and technically-involved, so it certainly isn’t meant for mainstream eyes. Still, it represents another avenue where Apple is being more open about its work.

At the end of last year, Apple said it would be open with academia and published its first AI research paper. This new online journal represents another step in that direction.

The debut post on ‘Improving the Realism of Synthetic Images’ has high production value, with rich formatting, GIFs, charts and other imagery for illustration to help explain the topic. There’s even a ‘Tips and Tricks’ section which provides some suggestions based on the experiences of the Apple engineers, so others can benefit.

The article does not detail how Apple went on to use the learnings in its own products and services, it is very much based around research and theory. Apple is open to feedback from interested parties; send questions and comments to machine-learning@apple.com.

Apple says that the site will be updated throughout the year; the posts are dated with month and year only, so do not expect extremely frequent updates like a typical news site.