Microsoft Pushes Its AI Strategy Forward with New Developer Tools

At Connect(); 2017, Microsoft today introduced a number of new tools for developers. The big focus this year is on artificial intelligence, and Microsoft is launching a new extension for Visual Studio that will help developers integrate AI into their apps.

Microsoft’s new Visual Studio Tools for AI extension lets users develop, debug, and deploy deep learning models in Visual Studio. The extension comes with support for the major machine learning frameworks, including Microsoft’s Cognitive Toolkit, Google TensorFlow, Caffe2, and MXNet. Using this new extension, developers will be able to develop deep learning models within Visual Studio, which can later be debugged right from the IDE itself. Developers would then be able to take advantage of the power of Azure Batch AI to train their deep learning models and deploy them in production using Azure Machine Learning. Visual Studio will also offer developers with a gallery of sample deep learning models which developers can play around with in order to build and test their models.

“With today’s intelligent cloud, emerging technologies like AI have the potential to change every facet of how we interact with the world. Developers are in the forefront of shaping that potential. Today at Connect(); we’re announcing new tools and services that help developers build applications and services for the AI-driven future, using the platforms, languages and collaboration tools they already know and love,” Microsoft’s EVP Scott Guthrie said in a statement.

Microsoft’s also bringing AI to embedded devices with the first ever beta release of Azure IoT Edge, which lets developers enable artificial intelligence and advanced analytics capabilities on the Internet of Things devices. Microsoft’s AI push is enormous for the company’s cloud business, and extensions like Visual Studio Tools for AI that simplify the engineering process for developers will be crucial when it comes to speeding up the adoption of AI.

Tagged with

Share post

Please check our Community Guidelines before commenting

Conversation 5 comments

  • IanYates82

    Premium Member
    15 November, 2017 - 5:00 pm

    <p>Interesting. I'll give this a look over the xmas holidays. If it's got some of the simple things sorted then it could allow for some fun possibilities. I write medical software for a living so I don't think I'll get to use this for work (that sort of thing would need a *lot* of testing, regulation, approval, etc) but maybe an IoT thing around the house could be neat.</p>

  • Winner

    15 November, 2017 - 6:36 pm

    <p>Just like Hololens pushed AR forward, and Windows 10 pushed UWP and the Windows Store forward, and XBoxOne pushed Kinect forward.</p><p><br></p><p>I just get the feeling like this is yet another push for the Next Big Thing that won't go very far. Hopefully this time I am wrong.</p>

  • skane2600

    15 November, 2017 - 10:59 pm

    <p>I admit I have no experience using any of these tools, but I find myself wondering if they're adequate for developing brand-new AI capabilities or just for "mining" well-known areas like speech-recognition. I have some very dated experience using Neural Nets to develop a medical product and my experience suggests that the access to tools is the easiest part. BTW, the medical product failed to live up to expectations.</p>

  • MikeGalos

    16 November, 2017 - 10:11 am

    <p>A full suite of AI tools fully integrated across the amazingly flexible set of Microsoft's cross-platform dev tools that support everything from IoT to Edge to Cloud. Wow. Nice job at moving the industry forward.</p>

  • Jules Wombat

    16 November, 2017 - 1:34 pm

    <p>I can see what they are attempting to do here. It all looks nice and integrated as we would have expected from an integrated Microsoft mindset. But using Visual Studio Python is much too slow for the heavy grunt required for serious Deep Learning Algorithms. A Visual Studio Pyhthon session is some 5 x slower than the native Python on a Keras 5 layer network, which everyone now deploys onto Graphcis card by Google Tesor Flow. That is simply an unacceptable benchmark for an AI developer, when a typical convolutional model takes some 4 hours to approach convergence performance. </p><p><br></p><p>Most AI developers need heavy grunt of native python, with the ease of an Keras API, Google Tensor Flow and CuDNN running on powerful Graphics cards. Microsoft is again just playing micky mouse tools in an area they really cannot catch up in. </p>


Stay up to date with the latest tech news from!

"*" indicates required fields

This field is for validation purposes and should be left unchanged.

Thurrott © 2023 BWW Media Group