The open source drive continues for Microsoft, as the company has announced that it is open sourcing the ONNX runtime engine that lies at the heart of its Windows machine learning platform.
Or rather, Windows Machine Learning platform.
The ONNX engine is a key part of Windows ML, with the Seattle based company having built this machine learning interface into Windows 10 to entice developers to use trained machine learning models in their apps for the operating system.
This basically enables the inference engine to evaluate trained models locally on Windows devices, instead of requiring the developers to use of the cloud.
Microsoft has already forged an alliance with Facebook around the Open Neural Network Exchange, with the two companies announcing the format in 2017 to make it easy for developers to move deep learning models between different frameworks.
Even though Microsoft still has and maintains its own Cognitive Toolkit, but it has recently started to downplay it in favor of Facebook PyTorch and Google TensorFlow.
This is what the company said recently:
“Microsoft believes an open ecosystem will help bring AI to everyone. We’re seeing traction for ONNX and Python with developers and data scientists so we are increasing our investments in those areas, while we continue to support Microsoft Cognitive Toolkit. We have nothing else to share at this time.”
This traction may well pick up pace now that ONNX is open source on Windows, and available for developers to customize and integrate the runtime into their existing systems to compile and build it for a variety of operating systems.
Microsoft officially confirmed the open sourcing at its Connect(); 2018 event, which also paid witness to Azure Machine Learning hitting general availability.
Big move from the company this, hopefully one that draws attention to the solution and drives its acceptance among developers — particularly now that the platform is out of the preview with the Windows 10 October 2018 Update.
If you’d like to get started with Windows Machine Learning, GitHub is the perfect place.