Slashdot Mirror


Is Microsoft Mainstreaming Machine Learning? (networkworld.com)

Tuesday Microsoft updated their open source Microsoft Cognitive Toolkit (CNTK), adding support for both C++ and Python. "This announcement is more than a point release..." argues Network World. "It's the recognition of AI and machine learning as the next big platform after mobile." This announcement represents a shift in Microsoft's customer focus from research to implementation... The toolkit is a supervised machine learning system in the same category of other open-source projects such as Tensorflow, Caffe and Torch. Microsoft is one of the leading investors in and contributors to the open machine learning software and research community. A glance at the Neural Information Processing Systems conference reveals that there are just four major technology companies committed to moving the field of neural networks forward: Microsoft, Google, Facebook and IBM.
A Microsoft engineer described CNTK as "democratizing AI," according to Microsoft's announcement, which also notes that their toolkit "has been optimized to best take advantage of the NVIDIA hardware and Azure networking capabilities that are part of the Azure offering."

6 of 51 comments (clear)

  1. No. by Gravis+Zero · · Score: 5, Informative

    TensorFlow is what everyone is using because it works well and it has a nice license to go with it. Besides, willingly becoming reliant on anything Microsoft makes is a devil's bargain.

    --
    Anons need not reply. Questions end with a question mark.
    1. Re:No. by somenickname · · Score: 4, Informative

      TensorFlow is what everyone is using because it works well and it has a nice license to go with it. Besides, willingly becoming reliant on anything Microsoft makes is a devil's bargain.

      Logged in to write exactly this. I've worked on a number of projects in the past that could have been simplified by TensorFlow. I have an equal level of disdain for Microsoft and Google but, TensorFlow is pretty darn cool. And the licensing alone makes it better than anything that Microsoft would be willing to release. So, no, Microsoft isn't mainstreaming machine learning. These days, I don't think Microsoft is mainstreaming anything except surveillance and shitty user interfaces.

  2. one could argue... by e432776 · · Score: 2

    that this would not be unprecedented. After all, Microsoft was an absolutely essential player in "mainstreaming" personal computing. For better or worse, before PC "clones" arrived on the scene running MS-DOS and then Windows, computing was *very* expensive and not mainstream at all. This is in no way to defend other business practices MS has had over the years, but a careful look at the record should show that MS was key to the "PC revolution".

    I'll show myself the door now, thanks.

  3. Re:Hipster MBAs writing the headlines at Slashdot? by ShanghaiBill · · Score: 2

    Seriously, mainstreaming? That's a word now?

    "Mainstream" has been used as a verb since at least 1975, when the IDEA act was passed, and handicapped children were "mainstreamed" into regular classrooms. The meaning of the word is obvious, and the usage in the headline is cromulent.

  4. Re:no, not "machine learning" by ShanghaiBill · · Score: 2

    neural network models are prohibitively hard to interpret and draw conclusions from)

    Biological neural networks are also hard to predict and interpret, so I don't think that is a very good argument.
    A system doesn't become less intelligent just because you don't understand how it works.

    Deep NNs have become a "fad" because for many applications, they work better (sometimes far better) than the alternatives.

    If you disagree, perhaps we can settle this over a nice game of Go.

  5. There is so much more to ML than a code library. by ZecretZquirrel · · Score: 3, Insightful

    It's like thinking that having a compiler means that I can program.