Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
AI Cloud Microsoft Open Source

Is Microsoft Mainstreaming Machine Learning? (networkworld.com) 51

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."
This discussion has been archived. No new comments can be posted.

Is Microsoft Mainstreaming Machine Learning?

Comments Filter:
  • No. (Score:5, Informative)

    by Gravis Zero ( 934156 ) on Saturday October 29, 2016 @03:59PM (#53176243)

    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.

    • by Anonymous Coward

      Might be good for a sexbot though. I think Tay may have broken records for AI learning.

    • Re:No. (Score:5, Informative)

      by somenickname ( 1270442 ) on Saturday October 29, 2016 @04:07PM (#53176275)

      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.

      • There's also Torch7, for example. Why should a Microsoft release, of all things, be so memorable?
      • by Anonymous Coward

        You do realize that the license is available with the toolkit and can actually be read instead of having to speculate on what is in it? Here it is in case it's too much trouble for you to look it up yourself:

        Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sel

        • by Anonymous Coward

          IAAL. This is the MIT License, a straight-up attribution-style license with a warranty disclaimer, about as free and OSS a license as you can get.

    • The question though is "everyone" that you refer to an academic or niche machine learning user aka not the "Mainstream".

      If you said "All of the CDC statisticians I know use linux" that doesn't logically follow that "Linux is mainstream".

      The implied argument is that existing machine learning is only being used by a select few, while Microsoft hopes to expand the market vastly beyond the current user set. Everyone could be using a Mainframe... and yet Apple could still mainstream computer usage with the rel

      • The question though is "everyone" that you refer to an academic or niche machine learning user aka not the "Mainstream".

        You are right, that is the question. The answer is that it's much more than academics or niche users... and it's for the same reasons.

    • TensorFlow is what everyone is using because it works well and it has a nice license to go with it.

      Can TensorFlow run on Windows?

      • Yes. I have been using tensorflow using something called 'bash on ubuntu on windows' which is a bit like Wine. The whole thing was very convenient and everything that I needed could be installed with apt-get.
    • by ranton ( 36917 )

      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.

      Oh yes, because willingly becoming reliant on anything Google has never led anyone astray. Google's history of ongoing support for their products is legendary.

      • The license (APL) is such that Google does not control the source code. Now that it's out, there's no going back. Microsoft's license on there other hand has some caveats.

  • Deep learning does not constitute all of, or even a significant chunk of, machine learning. It is merely the latest fad, much like personalized medicine in medical informatics or nano-machines in biochemistry.

    (I'd even go as far as to say that it's one of the worst parts of the field since neural network models are prohibitively hard to interpret and draw conclusions from)

    • 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 [wikipedia.org].

    • by gweihir ( 88907 )

      People that go for fads like this one usually think they are "cool", "modern" or even "cutting edge", when nothing like that is true. While I fully agree with your statement, those people cannot even understand what you are saying.

      • People that go for fads like this one usually think they are "cool", "modern" or even "cutting edge", when nothing like that is true.

        Exactly. The really cool kids, like you, just sit on the sidelines and sneer at the dorky nerds that are getting stuff done.

  • 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.
    • IBM AI: Win at jeopardy.
      Google AI: self-driving cars + winning at Go.
      DARPA AI: self-piloting drones.
      Microsoft AI: Racist sexbots [wikipedia.org]
    • 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.

      That is much more attributable to Intel and PC manufacturers, rather than to Microsoft.

  • if you want to make racist twitterbots. ;)

  • by ZecretZquirrel ( 610310 ) on Saturday October 29, 2016 @06:00PM (#53176629)
    It's like thinking that having a compiler means that I can program.

Some people claim that the UNIX learning curve is steep, but at least you only have to climb it once.

Working...