Prior to its build conference, Microsoft today released a number of new machine learning products and enhancements to some of its existing services. These range from no-code tools to hosted notebooks with a range of new APIs and other services. The core issue here, however, is that Microsoft continues its strategy of democratizing access to AI.
Prior to release, I met with Eric Boyd, Deputy CEO of Microsoft's AI platform, to discuss Microsoft Take up this space by competing heavily with Google and AWS and numerous, often specialized startups. To a certain extent, the actual machine learning technologies have become table settings. Everyone now offers pre-learned models, open source tools and training, modeling and deployment platforms. If a company does not have pre-formed models for some use cases that its competitors support, it is only a matter of time before they do. It is the support and experience of the developers that companies like Microsoft can differentiate with their long development of these tools.
"AI affects the way the world does business," Boyd said. "We see that 75% of business enterprises are doing more with AI in the next few years. According to Gartner, it has tripled in recent years. So we really see an explosion in the amount of work that comes from there. As people push this as businesses move forward, developers are at the forefront of trying to figure out how to move their businesses forward, how to build these models, how to develop those applications, and how to help make scaling changes to move through this development. "
What these companies – and their developers – need are more powerful tools to help them be more productive and build their models faster. At Microsoft, where these companies are often large companies, this also includes the ability to meet the needs of a business and provide the security guarantees they need.
However, as companies begin to learn machine learning, they are increasingly becoming to the point where they have moved from a few tests to a hundred models in production. This brings its own challenges. "They're trying to figure out how to control the lifecycle of these models," he said. "How do I think about the operating cycle? How do I think about a new model that I can provide? When is it ready to go? "
A few years ago, the industry started using a DevOps model for managing code. Essentially, Microsoft wants to switch to MLOps to manage models. "It's very similar to DevOps, but there are some differences in how the tools work," said Boyd. "At Microsoft, we really focus on how we can solve these issues to make developers more productive by using these enterprise tools to drive the necessary changes in their business." This means considering how concepts like source control are introduced can and the constant advancement of machine learning models, and this requires new tools.
It is no surprise then that adding more MLOps features is a big part of today's releases. For example, the company integrates some of these features into Azure DevOps so they can trigger release pipelines. The company also provides developers and data scientists with model version control tools, such as tracking and managing their assets, and sharing machine learning pipelines.
However, these are very many tools for advanced users of machine learning. On the other side of the spectrum, Microsoft announced a number of automated machine learning tools, including one that essentially automates all processes, and a visual model maker that emerged from the Azure ML Studio. As Boyd told me, even companies like British Petroleum and Oregon's Deschutes Brewery (try their Black Butte Porter if you have a chance) use these tools now.
"We've added a number of features to automated machine learning to simplify the process People are trying to use this kind of work," said Boyd.
Microsoft also has a number of new services in the cognitive field today Services, including a new personalization service, a handwriting recognition API, and another to transcribe conversations with multiple speakers.The personalization service is characterized by its use of enhanced learning, a different method of machine learning than most other cognitive ones For business users, there's also the Form Recognizer, which makes it easy to extract data from forms.
More interesting than the specific features is that Microsoft's focus here is one of more interesting little shifted. "We bew We move away from some first-level problems of "Here are the table inserts, you must have an AI platform" to much more complex use cases in terms of the operations of these algorithms, their simplification and new user experience to the operation of developers really simplify, and much better cognitive services, "Boyd explained.