Qi Lu, Microsoft’s executive vice president of Applications and Services, introduced the concept of Open Mind Studio in a talk titled “Machine Learning @ Microsoft” at the Stanford Scaled Machine Learning Conference earlier this month. The talk was highlighted here and here.
Although Microsoft would not comment on the platform beyond Lu’s slides, it appears that Open Mind Studio will consist of data, model, algorithm, pipeline, experiment and life cycle management components to help provide programming abstractions for machine learning.
Machine learning and artificial intelligence in general have become top of mind for leading technology vendors, and Microsoft has been among the pioneers in researching new ways to apply machine learning to make computing easier and more efficient for users.
Indeed, machine learning has been pervasive across Microsoft products. Machine learning, or ML, has found its way into more and more Microsoft technologies, including Bing, Skype, Kinect, HoloLens and Windows Phone. Whenever you use the search engine Bing, you’re using many components that have been trained with machine learning. In addition, Microsoft is using machine learning in security. The company arms its malware analysts with machine learning-driven technology, both to give the analysts “superpowers” to make them much more effective at searching through lots of data and also by autonomously helping to find malware authors.
Microsoft has taken its own path in the quest to deliver cognitive computing technology to consumers, competing with the likes of IBM, Google, Amazon and others. Cognitive computing encompasses various forms of artificial intelligence (AI), including machine learning, reasoning, natural language processing, speech and vision, human-computer interaction, dialog and narrative generation, and more. However, cognitive computing is but a subset of AI.
According to Lu’s slides, the Open Mind Studio project will rely on a federated infrastructure of data storage, compliance, resource management, scheduling and deployment components. The platform will support Microsoft’s open-source deep learning toolkit, the Computational Network Toolkit (CNTK). According to Microsoft Research, CNTK is a unified computational network framework that describes deep neural networks as a series of computational steps via a directed graph. In a directed graph, each node represents an input value or a network parameter, and each edge represents a matrix operation upon its children.
CNTK provides algorithms to carry out both forward computation and gradient calculation, Microsoft said. Lu described CNTK as Microsoft’s open-source, cross-platform toolkit for learning and evaluating models, especially deep neural networks.
Open Mind Studio also will integrate with other non-Microsoft deep learning frameworks such as TensorFlow, Caffe, MxNet, Theano and Torch, as well as with open-source computation frameworks including Hadoop and Spark. In addition, the project will support specialized, optimized computation frameworks such as Microsoft’s SCOPE and ChaNa.
Microsoft Eyes Visual Studio-Like Toolset for Machine Learning
ChaNa is a remote direct memory access (RDMA) optimized computation framework that is aimed at providing faster network performance. Early results show an order of magnitude improvement in performance, Lu’s slides show. SCOPE, or Structured Computations Optimized for Parallel Execution, is a SQL-like language for scale-out data processing based on Dryad for optimizing joins. Dryad is a Microsoft technology that enables programmers to use the resources of a computer cluster or a data center for running data-parallel programs.
“If Open Mind works as advertised, Microsoft will be making great, well-organized progress in helping to simplify enormously complex machine learning and deep learning processes,” said Charles King, principal analyst at Pund-IT.
While that is clearly important for companies pursuing machine learning projects, making complex processes simple also helps organizations make these projects commonplace, he said.
“If AI is to become as massively important and impactful as its proponents believe, the ML/DL tools supporting it need to be accessible to and useful for a wide variety of individuals and organizations,” King noted. “That appears to be a fundamental goal for Microsoft’s Open Mind.”
Meanwhile, according to Lu’s slides, Microsoft also has been working to scale machine learning on new hardware, including field-programmable gate-arrays (FPGAs).
As noted above, Microsoft has a variety of machine learning technology in place, including Azure Machine Learning, which became generally available last year.
At its Build 2016 developer conference, Microsoft introduced a new cloud-based Cortana Intelligence Suite that introduces cognitive technology for developers to use in building systems. Microsoft also announced a preview of its new Bot Framework, which enables organizations to build intelligent agents, known as bots.
At the conference, Microsoft CEO Satya Nadella said this is a new era of conversational intelligence and Microsoft is attempting to facilitate to enable developers to create more personal computing for every customer, business and industry.
“It’s a simple concept, yet it’s very powerful in its impact,” Nadella said in his keynote. “It is about taking the power of human language and applying it more pervasively to all of our computing—and to infuse into our computing and our computers, intelligence about us and our context. By doing so, we think this can have as profound an impact as the previous platform shifts have had—whether it be GUI, whether it be the Web or touch or mobile.”