AI, ignite2022, announcements, microsoft, artificial intelligence, machine learning

Microsoft Ignite 2022 – AI Announcements

Microsoft goes back in person with Ignite 2022. After a couple of years of only virtual events, we are back at in-person. Correction! Hybrid!

Yes, Ignite 2022 is Hybrid meaning you can enjoy all the sessions from home or all the way in Seattle USA.

If you are reading this article before the 13th of October, you can still join it online if you register at

What is announced and will be announced?

As a Microsoft partner, we focus solely on the Microsoft AI stack, so therefore also our main interest is in events like this. Let’s go over the announcements. Most of these items are coming in Preview first. But most of the things that were announced during the previous event, are now in GA.

Azure Machine Learning registries

Registered models
Model Registry in Databricks, no example yet for ML Registries

Helps ML professionals promote, share, and discover ML artifacts such as models, pipelines, and environments across multiple workspaces in an organization, enabling customers to track model and data lineage across workspaces in different Azure subscriptions, making cross-team operations easier.

Learn more about this new service at

Azure Container for PyTorch and Azure Data Science VMs for PyTorch ©

Curated environments and custom images that bundle innovative technologies used by Microsoft to set up, develop, accelerate, and support optimized training for PyTorch models.

Learn more about these containers at

Responsible Al Dashboard

thumbnail image 4 of blog post titled 
							Responsible AI dashboard: A one-stop shop for operationalizing Responsible AI in practice
Part of the Responsible AI Dashboard. Showing an error analysis

The Responsible AI Dashboard is a real must-use for all data scientists. It has been in preview for a long time, but now it will get into GA very soon!

The Dashboard makes it possible for customers to implement Responsible Al more easily by debugging machine learning models and making informed data-driven decisions.

Learn more about Responsible AI Dashboarding at

Automated Machine Learning enhancements

Support for natural language processing and image tasks.

Learn more about NLP and image tasks with AutoML at

AutoML code generation across tabular, image and text data.
Learn more about code generation at

Python SDK v2

Simplifies managing the entire ML lifecycle from training single jobs to pipelines and model deployments with Python functions.

Apache Spark pools integration

Enables ML professionals to quickly iterate on data preparation at scale on Spark clusters within AzureML platform.

Azure Form Recognizer

Form Recognizer is a great service for document recognition. It gives you a studio where you can easily build your own models or use some of the prebuilt tools.

The good news, the prebuilt models will be extended with a specific model to recognize parties and payment terms. And they expanded language support to 275 languages!

Azure Open AI Service

Teddy bears working on new AI research on the moon in the 1980s

You might have heard of Dall.E already, the AI model that allows users to generate realistic, accurate images, using text or images. (See picture). Until now you had to deploy that model on your own environment. But Microsoft added Dall.E to his Open AI Services (Private Preview) meaning you can just use the model by doing a REST call.

Azure Cognitive Services for Language

If you visit Microsoft Azure Language Studio (Language Studio – Microsoft Azure) you will notice that a lot is already possible out of the box. But some improvements are on their way.
The summarization capabilities got expanded to be able to work with abstractive documents and conversations. And it will even recognize chapter segmentation
Language support got expanded to 90 different languages.

Learn more about the above at

Computer Vision

Spatial Analysis measures dwelltime in checkout queue
Spatial Analysis

Image Analysis 4.0, an updated model, extracts a wide variety of visual features from images to improve digital asset management and customer accessibility.

Spatial Analysis on the Edge improves safety and security by ingesting streaming video from cameras, extracting insights, and generating events to be used by other systems like buildings and automobiles.

Azure Cognitive Services for Speech

New languages and emotions added to Neural Text to Speech to improve and scale the reach of voice assistants and more

Synthetic voice training with Custom Neural Voice including multi-style tones like cheerful or sad

Embedded Speech, enabling companies to use Azure speech services on their devices

New customization features like lexicons to improve the accuracy of speech recognition

Improved audio processing of all Azure prebuilt voices


Stay tuned.