Make adoption of AI as easy as possible

Only a few years ago, artificial intelligence was largely relegated to universities and research laboratories, a charming little-used computer concept in the business world. Today, artificial intelligence is built into everything from your fridge to your favorite workout app.

"It's really exciting, because every month, or every week, a new breakthrough is decisive," said Lance Olson, director of program management for AI at Microsoft. "Increasingly, conversations are moving from the art to the opportunity to move on to the next stage of implementing a specific project."

Nevertheless, many companies are struggling to achieve their goals in artificial intelligence because the supply of scientists and data experts has failed to meet the growing demand . Creating AI models is hard work. And then comes a struggle to get them into production – and keep them going. Data ages much faster than code, making models less accurate as the world changes around us.

At its Microsoft Build 2019 conference, the company aims to help all developers, even those with no computer or IT background, use its tools and services to deliver the important benefits expected by more in addition to customers.

"Artificial intelligence and machine learning can turn developers into heroes, as they offer customers truly personalized and super-immersive experiences," said Wisam Hirzalla, director of operational databases and product marketing. Blockchain at Microsoft. "We want to make it easier for businesses to use this technology."

Simplified and automated automatic learning

To this end, Microsoft is announcing new features for its cloud-based Azure Machine Learning service to enable developers and IT professionals of all skill levels to create advanced machine learning models.

We can think of AI practitioners in three categories, according to Bharat Sandhu, director of artificial intelligence at Microsoft. First, we have developers and scientists who like to write code. They want to build machine learning models with the help of tools and processes that they already know. For them, Azure Machine Learning offers a "first code model," in which they can use the development tools that they like.

A second group, including business experts, may know a lot about data, but not about machine learning or code. Azure Machine Learning automated learning machine learning is a "no code" option, accessible without the need to write code.

"A third category of people, who are learning the concepts of machine learning, want to create their own models, but they are not coders. It can be from IT professionals or people with a background in statistics or mathematics, "said Sandhu. "For these customers, we offer a drag-and-drop experience to create models visually."

Sandhu noted that no matter how machine learning models are created, they all use the same back-end, which means that all models can easily be integrated together.

L & # 39; interoperability

Of course, developers and data specialists have a large number of platforms to choose from when building AI models. In order for companies to embrace AI advances as quickly as possible, Microsoft said it was important to address imbalances between platforms, which could delay the release of these models.

One of the ways Microsoft promotes interoperability between the various artificial intelligence environments is a standard called ONNX Runtime or Open Neural Network Exchange. This joint effort with other technology companies creates deployment models that work across multiple platforms.

This frees developers and data scientists from the most appropriate use of framework and hardware. And this allows the operational team to focus on deploying and getting results, instead of having to perform the translation as it goes from one to the other. ;other.

At the time of construction, Microsoft announced support for ONNX integration with leading hardware accelerators.

The company also announces that it is now actively contributing to the MLflow project, an open source platform for managing the lifecycle of machine learning.

Azure Cognitive Services Updates

More than 1.3 million developers, many of whom do not have specific AI or computer skills, currently use Azure Cognitive Services to create intelligent applications that can see, hear, speak, understand, and even to start reasoning.

When creating, Microsoft announces a new Azure Cognitive Services category called Decision, which provides specific recommendations to help users make decisions. This new category includes Personalizer, which uses an artificial intelligence branch called reinforcement learning to help technology glean insights from its own experiences and then make informed recommendations.

"We are able to follow reinforcement learning and deliver it in an accessible way to developers and do not need a computer scientist," said Olson. "It will be very impactful for the customers."

Upon creation, the company announces many other updates to Azure Cognitive Services, including Ink Recognizer, which can learn to read handwriting, Form Recognizer, which identifies forms, as well as other information. 39 other new conversational transcription features and other advances in speech, vision and language.

Just start

To date, Microsoft customers have created nearly 400,000 digital agents through its Azure bot service, and more than 3,000 are online each week. Businesses of all sizes are turning to AI to give them a competitive advantage.

This includes Cheetah Mobile, a leading mobile application manufacturer that builds enhanced hardware for artificial intelligence, including the portable CM Translator. Rather than fully developing the voice system, the company used Azure Cognitive Services to leverage its text-to-speech API to provide fast, high-quality translations.

Development cost savings have kept the device affordable, without compromising natural speech flow.

Other companies claim that one of the key benefits of using Azure's data and AI tools is that they can take advantage of other attributes built into these tools. For example, MediaValet, a digital asset management company, relies on the security and privacy assurances provided by Azure to reassure its customers that the images it processes will be processed properly. .

"We're not a big company, but we can play football with big business because we can leverage the security and privacy attributes of Azure's trustworthiness," said Jean Lozano, director of MediaValet technology.

In the coming months and years, Microsoft expects more and more customers to start using AI, both because they see the benefits for the company. business and because the tools are more accessible.

"Artificial intelligence opens up many possibilities. And the limits are very few, usually limited only by your imagination, "said Olson. "It does not have to be overwhelming for people. We are getting to the point where we can now make AI accessible to a much wider group of customers. "

In connection with the news of AI at Microsoft Build 2019: