Google’s Cloud AutoML Vision can automatically create an accurate machine-learning training model using relatively small amounts of data. (Google Photo)

Google has figured out a way to make sophisticated machine-learning techniques available to cloud customers who want to use that technology in their apps without having to hire an expensive machine-learning team.

The new set of services is called Cloud AutoML, and later on Wednesday Google is expected to announce that it’s ready to open the service up to a select number of users. Cloud AutoML uses customer data — even small amounts — to automatically generate a machine learning model best suited for that data, and produce results.

“At Google Cloud AI, our goal was to lower the barrier of entry and make AI possible to the the largest possible community of developers, researchers, and businesses,” said Fei Fei Li, chief scientist at Google Cloud AI and director of the Stanford Artificial Intelligence Lab, in a press conference on Tuesday outlining AutoML.

(L to R): Jia Li, head of research and development for Google Cloud AI; Fei-Fei Li, chief scientist, Google Cloud AI (Photo via Twitter)

Unless you’ve got a machine learning expert on the payroll, using those techniques to build the appropriate training model for your data set is extremely difficult. And those folks don’t come cheap; demand for AI experts is off the charts right now, and with a relatively few number of qualified people that can fill those jobs, a disproportionate amount of machine-learning talent is concentrated at the big platform companies.

The first version of the services released Wednesday, Cloud AutoML Vision, gives users a relatively simple drop-down menu from which to upload image data and provide a description or label for that data. The key is that AutoML Vision will automatically train a machine-learning model for that data using well-known but complex techniques like learning to learn and transfer learning, said Jia Li, head of research and development for Cloud AI.

A demonstration of how Google’s Cloud AutoML works. (Google Image)

Cloud AutoML Vision will help users identify unlabeled images within a database of images when given a correctly labeled set of images to think over. In some cases, they might only need to upload dozens of images, Jia Li said, which gets around the other major AI accessibility issue: the large datasets needed to accurately train most models also don’t exist outside the major tech platform companies.

Google said the results from Cloud AutoML are more accurate than other “generic ML APIs,” but it declined to get into details. Retail and healthcare companies have been testing Cloud AutoML to automatically classify products or find very specific images, said senior director of artificial intelligence for Google Cloud AI.

As it looks up at rivals Amazon Web Services and Microsoft in the cloud computing market, Google has focused on its AI research as a competitive advantage over the larger companies. AWS announced an easy-to-use machine-learning service called Sagemaker at re:Invent 2017, but isn’t clear how much overlap there is between the two services.

You’ll have to apply to use Cloud AutoML on Google’s site, but wider availability and other Cloud AutoML services are coming, Google said.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.