Machine learning is still in its infancy, and Microsoft plans to help that along through the release of a new suite. The information comes out of the Stanford Scaled Machine Learning conference, where Microsoft’s Qi Lu announced “Open Mind Studio.”
Thanks go once more to Walking Cat on Twitter, who discovered the above slide. Microsoft already offers the Cortana Intelligence Suite as one such suite, as well as their Azure Machine Learning Studio. Open Mind will go one step further, however.
A Variety of Frameworks and Platforms
Lu’s slides reveal that Open Mind Studio will run on a variety of computing platforms, including:
- CPU
- GPU
- FPGA
- RDMA
- Cloud
- Client/ Device
It also appears that the platform will use a federated infrastructure for data storage, compliance, resource management, scheduling, and deployment. However, perhaps the biggest difference is the range of third party frameworks available.
As well as CNTK, Open Mind will support deep learning frameworks such as Caffe, MxNet, TensorFlow, Theano and Torch. It will also utilize open source computation frameworks like Hadoop and Spark.
Additionally, Microsoft will support SCOPE and ChaNa, their specialized, optimized computation frameworks. ChaNa is the company’s personal RDMA computation framework which is focused on faster networking. SCOPE, on the other hand, is Microsoft’s language for Cosmos, which is used for their big data computing.
Lu also talked about the history of machine learning and their future plans. Their use of the technique first started in 1997 with the launch of Hotmail and its junk filter. Since then it’s been implemented in Bing Maps and Search, Kinect, Skype and more.
In the future, the company will continue to embed the technique in Office and Dynamic 365, Skype, Bing, and Cortana. Microsoft also wants to build it into tablets and phones, as well as the Azure infrastructure.
This is the first we’ve heard of Open Mind, and Microsoft has refused to give any further information. However, it could be that the official reveal will be at the Data Science Summit in September, so we’ll keep you updated.
You can view Lu’s slides for yourself here.
