Prior to its construction conference, Microsoft today released a series of new machine learning products and tweaks to some of its existing services. These range from non-coding tools to hosted notebooks, with a host of new APIs and other services in between. However, the core theme here is that Microsoft continues its strategy of democratizing access to AI.
Before the release, I sat down with Microsoft's Eric Boyd, the company's vice president of the AI platform, to discuss Microsoft's take on this space, where it competes heavily with Google and AWS, as well as many, often more specialized startups. . And to some extent, the actual machine learning technologies have become table games. All now offer pre-trained models, open source tools and platforms for training, building and distributing models. If one company does not have pre-trained models for any uses that its competitors support, it is just a matter of time before it will. There are additional services and the overall developer experience, where companies such as Microsoft, with their long history of developing these tools, can distinguish themselves.
"AI really affects the way the world does business., Sa Boyd." We see that 75% of commercial businesses are doing more with AI over the next few years. It has tripled in the last couple of years, according to Gartner. And then we really see an explosion in the amount of work coming from there. As people drive this forward, as businesses drive this forward, developers are on the front line, trying to figure out how to move their businesses forward, how to build those models, and how to build those applications, and scale with all the changes that are moving through this. "
What these companies – and their developers – need is more powerful tools that allow them to become more productive and build their models faster. In Microsoft, where these companies are often large companies, it also includes being able to meet the needs of An enterprise and offering the security guarantees they need.
As companies start adopting machine learning, they now also realize to a point where they have moved from a few tests to perhaps running a hundred models in production, with their own challenges. "They are trying to figure out how to handle the life cycle of these models," he said. "How do I think about the cycle of operations? How do I think about a new model that I'm ready to distribute? When is it ready to go? "
Just a few years ago, the industry began moving to a DevOps model to manage code. What Microsoft essentially wants to move to is MLOps for managing models." It's very similar to DevOps, but it is some different differences in how the tools work, "Boyd said." In Microsoft, we focus on how we solve these issues to make developers more productive, using these business tools to drive those changes they need across the organization. "This means thinking about how to include terms such as source control and continuous development into machine learning models, for example, and it will take new tools.
Adding more MLOps features is not a big part of today's The company integrates some of these features into Azure DevOps, for example, allowing them to trigger release pipelines, and also provides developers and computer model modeling control tools such as tracking and managing their assets and sharing machine learning pipelines.
However, this is very much a tool for advanced machine learning practitioners. On the other side of the spectrum, Microsoft also announced a number of automated machine learning tools, including one that mainly automates all processes, as well as a visual model builder that grew out of Azure ML Studio. As Boyd told me, even companies like British Petroleum and Oregon use Deschutes Brewery (try Black Butte Porter if you get a chance) reach these tools.
"We've put a lot of features into automated machine learning to simplify how people try to use this type of work," Boyd said.
Today, Microsoft launched a number of new services in its Cognitive Services lineup, including a new personalization service, an API for recognizing handwriting, and another for transcribing conversations with multiple speakers. The personal service is out here because it uses reinforcement learning, another machine learning technique from most other cognitive services, and because it is far easier to implement than similar services. For corporate users, Form Recognizer also makes it easy to retrieve data from forms.
It is even more interesting that the specific features are that Microsoft changes its weight a little. "We are moving away from some of the first level issues of" here's the table games, you have to have an AI platform, "to much more sophisticated usage cases around the operations of these algorithms, simplifying them, new user experiences to really simplify how developers jobs and much richer cognitive services, "Boyd explains.