Overview
- Microsoft has announced the release of Windows ML for Windows 10
- You can bring your ML models directly to your desktop apps
- It will be launched with the next major Windows 10 update
Introduction
Microsoft wants to help developers bring their machine learning models from the cloud to their desktop. The AI platform will be launched with the next major update to Windows 10.
Rather than taking a circuitous route to the cloud and back, the platform will make direct use of the GPU on your machine so you can run your models in real time. Microsoft has been talking up their machine learning capabilities in the cloud since a long time and with this addition, it completes Satya Nadella’s vision for their AI journey.
Developers can now build their models in the cloud and integrate them in their desktop applications using Visual Studio and other tools.
The technology behind the AI platform
A project named Onnx, backed by the big tech giants – Microsoft, Facebook and Amazon, is at the heart of this technology. Developers can convert PyTorch, Caffe2 and other models into the Onnx format and move them between different frameworks.
Keeping up with the recent trend of automated machine learning tools, Microsoft will also give developers full access to train image recognition models on it’s Azure Custom Vision Service. The models can then seamlessly be exported to WindowsML. The developer doesn’t need to be an advanced machien learning expert to do this – the tool just requires the training data to be properly tagged for it to work.
You can read Microsoft’s full announcement in their blog post here.
Our take on this
Microsoft was already using it’s AI capabilities in their Office 365 suite and their Photos app. We expect the data science community to benefit directly from this release. A lot of the popular machine learning tools and resources are often not found on the Windows platform or are late in arriving, which has been a drawback for data science practitioners.
I recently read a comment on LinkedIn from a data science thought leader urging people to forget about Windows when it comes to ML. With this release, Microsoft will hope to make it’s penetration into the market again and win over the naysayers.
Subscribe to AVBytes here to get regular data science, machine learning and AI updates in your inbox!
Senior Editor at Analytics Vidhya.Data visualization practitioner who loves reading and delving deeper into the data science and machine learning arts. Always looking for new ways to improve processes using ML and AI.
I think if microsoft can make use of the non Nividia GPUs, then it will make a huge impact, as lots of us have machines with almost useless GPUs that we wish to use in our models :) Great Article 👍
it would be great if Microsoft can add support for AMD GPUs through some sort of an API or whatever, if they did that i think they will gain a huge popularity in no time, as lots of us are frustrated by the fact that our GPUs are totally useless when it come to DL-models. Thanks for the article, keep up the good work ^^