Microsoft is outlining its new AI platform for Windows 10 app developers ahead of Build 2018 today. The company is bringing its expertise in artificial intelligence and machine learning to Windows developers in order to help them deliver more intelligent and advanced experiences.
The software giant is making it easier for developers to use the Open Neural Network Exchange (ONNX) in Windows applications by integrating it into Visual Studio. A future version of Visual Studio will allow developers to automatically general model interfaces for their machine learning models by simply adding the ONNX file to their Windows app project. Developers can utilize on-device machine learning by training these models on the cloud using the Azure Machine Learning Workbench and exporting them as ONNX models for use on their Windows applications.
In addition to ONNX support, Microsoft emphasised on some of the other benefits available to developers using its AI platform. Microsoft says the company’s ML model evaluation is optimized for modern hardware that allows developers to maximize performance, mostly thanks to optimized instruction sets for modern chips, hardware acceleration on GPUs with DirectX 12 support, and more built into Windows 10 with the next major Windows 10 update. Developers can also use Microsoft’s AI platform in the cloud to reduce costs by making use of the cloud-based services only when necessary and using on-device processing for faster response times in other use cases.
The software giant is also working to deliver superior performance by building support for AI co-processors. The company says it’s new AI platform isn’t only limited to PCs, and developers can use it across all types of Windows hardware ranging from the HoloLens to IoT edge devices. Redmond says this is just the beginning and hopes to share more at Build 2018 this May.