Much ado about what to do: Microsoft wants Azure to help decide
- 03 May, 2019 07:00
Microsoft has used its Build developer conference to announce a new Azure Cognitive Services category: ‘Decision’. The category groups together Azure’s Content Moderator and Anomaly Detector services, as well as a new service called Personalizer, which is currently in preview.
Personalizer employs reinforcement learning to help deliver recommendations tailored to an individual user and “provide a more engaging experience,” Jason Zander, head of global engineering Microsoft Azure, told a press briefing ahead of today’s official announcement.
Reinforcement learning “uses more and more and more decisions as part of the training, to get it to be better and better and better over time versus generic algorithms,” Zander said.
He said that ecommerce services were an obvious example of where it would prove useful, but is not the only use case for Personalizer.
“We are able to take reinforcement learning and ship it in a way that’s accessible to developers and doesn't require a data scientist,” Lance Olson, director of program management for applied AI at Microsoft, said in a statement. “That will be very impactful for customers.”
Cognitive Services’ vision category is also getting two new services, Frank Shaw, Microsoft’s corporate vice president of communications, told a media briefing.
Ink Recognizer offers digital ink recognition capabilities and Forms Recognizer allows text, key value pairs, and tables to be extracted from documents. Microsoft also detailed a new speech-to-text service called Conversation Transcription.
The company announced general availability of “cognitive search” capability in Azure Search.
It means that customers of Microsoft’s cloud services can “apply AI to extract insights and structured information from their content,” Shaw said. “In addition, Microsoft is extending the capabilities of cognitive search to apply to any variety of scenarios such as Power BI visualisations.”
Metropolitan Museum of Art, contract management company Icertis, and global engineering firm Howden are among the organisations already employing cognitive search on Azure.
Microsoft also unveiled a range of services intended to make it easier for companies to use AI and machine learning technologies, including a visual ‘zero code’ drag-and-drop interface for building and deploying ML models and ‘MLOps’ capabilities
“With MLOps, we are bringing the best of the DevOps world to machine learning to enable debugging and deployment, model validation, profiling, monitoring, and CI/CD leading to efficiency and improved model quality over time,” Shaw said.
Other Build announcements focused on Azure Blockchain Service, which is now available in preview, and ‘IoT Plug and Play’.
IoT Plug and Play is a “new open modeling language to connect IoT devices to the cloud seamlessly, enabling developers to navigate one of the biggest challenges they face — deploying IoT solutions at scale,” Scott Guthrie, EVP, Microsoft Cloud and AI Group, wrote in a blog entry.
“Previously, software had to be written specifically for the connected device it supported, limiting the scale of IoT deployments. IoT Plug and Play provides developers with a faster way to build IoT devices and will provide customers with a large ecosystem of partner-certified devices that can work with any IoT solution.”