Microsoft brings new brains to Azure AI at Ignite conference
Its automated machine learning service is designed to steer artificial intelligence jobs away from potholes and dead ends.
- Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Microsoft researcher Nicolo Fusi shows data from the company's automated machine learning technology.
This might sound a bit meta, but Microsoft is applying new digital brains to cut down on the difficulties of using artificial intelligence technology.
Artificial intelligence, which these days typically refers to technology called neural networks or machine learning modeled loosely on human brains, is moving from the exotic to the mainstream in the computing world. But that doesn't make it easy.
Which is why Microsoft, at its Ignite conference Monday, announced it's got a new service that tries to take some of the pain out of AI. Automated machine learning is designed to zero in on the best way to construct an AI model, whether that's for identifying what's in a video, analyzing medical scans, looking for manufacturing glitches or flagging fraudulent credit card transactions.
"It simplifies the work a data scientist has to do," said Eric Boyd, corporate vice president of Microsoft's AI platform work at its Azure cloud-computing division. "If you have a data set and need to learn a new model, there's a lot of tedium a data scientist has to do, iterating through algorithms and changing knobs for those algorithms."
Making AI easier is key to unlocking its benefits. It's spread to every corner of tech giants like Facebook, Google, Apple and Microsoft, but much of the promise of AI today is at ordinary companies that don't have legions of programmers and Ph.D.s on staff.
AI jobs are different, so Microsoft's new service is designed to try out lots of possible approaches, tweak the "hyperparameterrs" that govern how they run, and rapidly winnow out the dud approaches. The approach -- probabilistic matrix factorization if you're curious -- should help junior data scientists stretch their wings and make senior ones more productive, Boyd said.
Of course you'll have to pay to use Microsoft's automated machine learning, available on its Azure service. There, the more computing horsepower customers use, they more they have to pay Microsoft. But at least in principle, the expense is a better alternative to fumbling around manually to find the best approach for an AI job.
Cloud-computing companies like Microsoft, Google and IBM are scrambling to attract customers who need to run AI jobs on the fleets of servers that make up their data centers. AI spending should jump from $19 billion in 2018 to a whopping $52 billion in 2021, analyst firm IDC predicts.
For common situations like image recognition, Microsoft also offers customers pay-as-you-go access to its own AI models. At Ignite, the company announced another one for translating speech into text with nearly real-time speed. And it's expanded the number of AI services that'll run in accelerated form on programmable chips, called FPGAs, that are at the heart of the Microsoft's Brainwave AI work.
Taking It to Extremes: Mix insane situations -- erupting volcanoes, nuclear meltdowns, 30-foot waves -- with everyday tech. Here's what happens.
CNET Magazine: Check out a sample of the stories in CNET's newsstand edition.