Archive for the ‘AI Hardware’ Category

Microsoft Build Conference – 2018 – Part 1(AI)

June 7, 2018

Why did the Javascript developer need glasses? Because he didn’t  C# “ . Yes it is Build Time folks :)…

The Build conference  was held at Washington State Convention center was the home to the conference. Same as last year… It featured approximately 405+ sessions (Keynotes/deep dives/20 minute introductions sessions etc.) all packed in 3 days.

This year one of the things I was amazed by at Build was the healthy assortment of food and snacks. Very little high sugar/carb.  items and more veggies, fruits etc.

Also more healthier ambience in general like Performers -with a good mix of rock, new age, meditative, contemporary dance, and very unique….

Bummer L  I  missed out on the celebration especially the Silent disco.

Conference overview 

The theme of build was around   AI (intelligent apps), Azure IoT Edge, Windows 10 on ARM, Desktop apps modernization, Migrate apps. to cloud, Microsoft Graph and Teams …

Novelty items were Azure Sphere, Project brainwave, OpenAI framework (ONNX) support, ML.Net, Azure Databricks.

From the Keynotes Joe Belfiore mentioned that Microsoft wants developers to use data and build Intelligent apps and not associate themselves as .Net, Cloud Developers,… but consider themselves Microsoft 365 developers.

It was impressive to see  Satya Nadella was the only CEO who addressed 3 key pillars of responsibilities from  platform company. Google and FB didn’t address this in their own developer conferences!

For me one the words of wisdom in the Vision  keynote which suck a chord was from Satya Nadella —  “The time has come to ask ourselves not what computers can do  but what computers should do”…

AI for everyone … 

In this blog I cover primarily AI announcements and topics.

It was very clear that Intelligent apps are here to stay and will be democratized with Microsoft’s platform push for AI. This was evident as there were around 57 sessions dedicated to AI covering variety of topics:

  • New framework ML .Net, Open AI framework – ONNX
  • is open source
  • Azure Conversational AI (a.k.a bots)-100+ new features,
  • Smart Ink, Speech
  • Video AI – VideoIndexer,
  • Azure ML Vision,
  • AI for Security & AI for Accessibility,
  • Ethics


One of the Key use cases of AI is from customization to personalization…

Microsoft showcased their AI success across public AI benchmarks

New framework ML .Net, Open AI framework – ONNX

Microsoft is working with Facebook and other companies to create  open AI framework which allows to make use of multiple AI model architectures like CNTK, Tensorflow etc. and allow it to run on multiple hardware GPU’s/ FPGA’s `etc. This allows to decouple models from Hardware by proving a middle layer… ( open source)

This is great for all the .Net developers who don’t need to invest in another programming language. But this is more than a programming language. It is a framework.

Keras Framework –

Windows ML on Intel Platform

Intel is adding support for Machine Learning into the CPU, GPU and other Hardware architectures like FPGA’s and VPU’s.

Conversational AI a.k.a Bots

There have been more than a 100 features ad to the Bot framework.


Drawing Bots

At the core of Microsoft’s drawing bot is a technology known as a Generative Adversarial Network, or GAN. The network consists of two machine learning models, one that generates images from text descriptions and another, known as a discriminator, that uses text descriptions to judge the authenticity of generated images. The generator attempts to get fake pictures past the discriminator; the discriminator never wants to be fooled. Working together, the discriminator pushes the generator toward perfection.

Lessons learned from chatbot testing

Speech  – Unified speech SDK 

Speech services  and Cortana

Cortana Skills – Looks like the list is growing …

Video Indexer (inference 🙂 ) – Lots of great AI features

Intelligent ink – Allows cloud API service to decipher advanced ink. Windows already has built-in recognition of Ink and is already good at recognizing diagrams and free form handwriting. Intelligent Ink uses cloud APIs for ink Analysis.

sophisticated Ink analysis to find more semantic structure in writing

Finally the following slide explains why a cloud service for Ink rather than use the default local  Ink API

Microsoft office especially Powerpoint is leveraging this:

Ink to Shape

Transforms the picture below to come up with a nicer circle



Ink to text – It can also identify the text within a shape and convert that to a font. is a company doing much work  in this space

information on the session –

Azure ML computer vision – Toshiba Camera

Toshiba showcased their computer vision device which allows for detecting problems using Azure ML computer vision package.

Next time I will cover IoT – Azure IoT Edge, azure Sphere etc.



     All Build sessions