Making AI real for every developer and every organization

Gepost op 2 mei, 2019

Corporate Vice President, Azure AI

ai_background

AI is fueling the next wave of transformative innovations that will change the world. With Azure AI, our goal is to empower organizations to apply AI across the spectrum of their business to engage customers, empower employees, optimize operations and transform products. To make this a reality, we have three guiding investment principles:

  • Boost the productivity of developers and data scientists and empower them to build AI solutions faster.
  • Enable these AI solutions to be deployed at scale alongside existing systems and processes.
  • Ensure organizations can build with full confidence knowing that they own and control their data on a platform that adheres to some of the industry’s strictest privacy standards and has the most comprehensive compliance portfolio of any cloud provider.

These guiding principles enable us to fulfill our mission of empowering every developer and every organization to harness the potential of AI. With research centers that span the globe, from Redmond to Shanghai, we continue to achieve industry breakthroughs in areas such as vision, speech, language, advanced machine learning techniques, and specialized AI hardware. These innovations are now key components of several of our flagship products, like Office 365, Xbox, Bing and Dynamics 365. This is important because with Azure AI, customers can benefit from the latest innovations that have been thoroughly battle-tested in our own products.

We are honored and humbled by the tremendous adoption of Azure AI by customers. Organizations of all sizes in all industries are using Azure AI to transform their business by:

  • Using machine learning to build predictive models, optimizing business processes.
  • Utilizing advanced vision, speech, language, and decision-enabling capabilities to build AI powered apps and agents to deliver personalized and engaging experiences.
  • Applying knowledge mining to uncover latent insights from vast repositories of data.

Today, we are excited to announce a range of innovations across all of these areas. Let’s walk through them.

Machine learning

Azure Machine Learning service is designed to accelerate the end-to-end machine learning lifecycle. With Azure Machine Learning, developers and organizations can quickly and easily build, train, and deploy models anywhere from the intelligent cloud to the intelligent edge, as well as manage their models with integrated (CI/CD) tooling.

As we strive to enable developers, data scientists, and DevOps professionals across all skill levels to increase productivity, operationalize models at scale and innovate faster. We are pleased to announce:

New capabilities to enhance productivity now in preview:

  • Automated machine learning user interface that enables business domain experts to train machine learning models with just a few clicks.
  • Zero-code, visual interface that enables users new to machine learning to build, train, and deploy models easily using drag and drop capabilities.
  • Azure Machine Learning notebooks that provides developers and data scientists a code-first machine learning experience.

New capabilities to enable operationalization of models at scale:

  • MLOps or DevOps for machine learning capabilities, including Azure DevOps integration that enables Azure DevOps to be used to manage the entire machine learning lifecycle including model reproducibility, validation, deployment, and retraining.
  • General availability of hardware accelerated models that run on FPGA’s in Azure for extremely low latency and low-cost inferencing. Available in preview for Databox Edge.
  • Model interpretability capabilities that enable customers to understand how a model works and why it makes certain predictions, removing the ‘black box’ aspect of ML models.

Our commitment to an open platform:

  • Contribution to the open source MLflow project, with native support for MLflow in Azure Machine Learning service.
  • Support for ONNX Runtime for NVIDIA TensorRT and Intel nGraph for high speed inferencing on NVIDIA and Intel chipsets.
  • Preview of a new service, Azure Open Datasets, that helps customers improve machine learning model accuracy using rich, curated open data and reduce time normally spent on both data discovery and preparation.

It’s exciting to see customers such as BP, Walgreens Boots and Schneider Electric deploying machine learning solutions at scale using Azure Machine Learning.

“Using Azure Machine Learning service, we get peace of mind with automated machine learning, knowing that we are exhausting all the possible scenarios and using the best model for our inputs.” - Diana Kennedy, Vice President, Strategy, Architecture, and Planning, BP

Visit Azure Machine Learning to discover more.

AI apps and agents

The combination of Azure Cognitive Services and Azure Bot Service enables developers to easily infuse powerful AI capabilities into their apps and agents.

Azure Cognitive Services continues to be the most comprehensive portfolio in the market for developers who want to embed the ability to see, hear, respond, translate, reason and more into their apps. Today we’re making it even easier for developers to embed AI into their applications:

  • Introduction of a new Decision category.
    Services in this category provide users recommendations to enable informed and efficient decision-making. Services such as Content Moderator, the recently announced Anomaly Detector and a new service called Personalizer, available in preview, are part of this new category. Personalizer is built on reinforcement-learning and prioritizes relevant content and experiences for each user, improving app usability and engagement. Microsoft’s very own Xbox drove a 40 percent lift in user engagement on its home screen as a result of using Personalizer.
  • In Vision, we are announcing two new services available in preview. Ink Recognizer enables developers to combine the benefits of physical pen and paper with the best of the digital by embedding digital ink recognition capabilities into apps. Developers can build on top of it to make notes searchable and convert hand-written sketches into presentation-ready content in a matter of minutes. Additionally, the Computer Vision read capability, which extracts text from common file types including multi-page documents and PDF, TIFF formats, is now generally available.
  • In Speech, we are announcing preview of new advanced speech-to-text capability called conversation transcription that catalyzes meeting efficiency by transcribing conversations in real-time so participants can fully engage in the discussion, know who said what when, and quickly follow up on next steps. Neural text-to-speech capability and Speech Service Device SDK are also now generally available.
  • In Language, Language Understanding has a new analytics dashboard to evaluate the quality of language models. In addition, QnA Maker now supports multiturn dialogs. The Text Analytics named entity extraction capability is now generally available.
  • We have expanded the portfolio of Cognitive Services that can run locally through a Docker container and we’re pleased to preview container support for Anomaly Detector, Speech-to-Text, and Text-to-Speech.

Only Azure provides developers with the flexibility to embed these powerful AI services where needed. Visit Azure Cognitive Services to find out more.

Azure Bot Service, built on Microsoft Bot Framework, makes it easier to develop bots and intelligent agents. New enhancements include:

  • Adaptive dialogs enable developers to create more sophisticated, dynamic conversations.
  • Language generation package streamlines the creation of smart and dynamic bot responses.
  • Emulator now has improved fidelity for debugging channels.

LaLiga, the premier men’s soccer league of Spain, creates solutions using Cognitive Services, Bot Service, and other Azure services. Their intelligent bot gives LaLiga new ways to stay engaged with their fans on their preferred social platforms. Delivering a world-class voice assistant is key to scoring brand love with fans:

“Our digital innovation platform built on Microsoft Azure helps us deliver the best possible fan experiences for the world’s best sports league.” - Jose Carlos Franco, Head of Data and Analytics, LaLiga

Knowledge mining

While organizations have seemingly unlimited access to information that can range from databases to PDFs to media files, there are still significant challenges in making that information usable and meaningful. With knowledge mining, you can leverage industry leading AI capabilities to easily unlock latent insights from all your content at scale.

We have two exciting announcements in this category:

  • The cognitive search capability of Azure Search, is now generally available and up to 30 times faster than before. Azure Search is the only offering in the market with a single mechanism to apply AI enrichments to content. Using Cognitive Search and its built-in AI capabilities, customers can discover patterns and relationships in their content, understand sentiment, extract key phrases and more, all without any data science expertise. In addition, a new knowledge store capability in preview enables developers to further leverage the insights and metadata they extract from the cognitive search pipeline. Developers can store the enriched metadata they create with cognitive search and apply it to any variety of scenarios such as Power BI visualizations, custom knowledge graphs, trigger actions within an application, or build machine learning models with new labeled data.
  • The new Form Recognizer service applies advanced machine learning to accurately extract text, key-value pairs, and tables from documents. With just a few samples, it tailors its understanding to supplied documents, both on-premises and in the cloud. It can be used to build robotic process automation (RPA) solutions.

The Metropolitan Museum of Art is exploring how Cognitive Search understands nuances and relationships across their encyclopedic collection:

British Petroleum, Icertis, Howden, Chevron, UiPath, and others benefit from Azure Search and Form Recognizer to extract insights from their content and automate processes.

“We’re excited to leverage Form Recognizer as a key document extraction capability on our Robotic Process Automation (RPA) platform and our open AI ecosystem. UiPath’s and Microsoft’s investments in AI are streamlining the process of unlocking key business data, making possible a new era of AI-driven business insights and knowledge management” - Mark Benyovszky, Director Artificial Intelligence, UiPath

Continuing to innovate

We continue to invest to make Azure the best place for AI and we are most excited to see how customers are applying AI in their businesses. The opportunities are limitless, and we are looking forward to seeing what you create with Azure AI.

Additional resources

Ready to get started? Check out: