Microsoft Open Sources Its Machine Learning Toolkit (thestack.com)
An anonymous reader writes: Microsoft has this week made its Distributed Machine Learning Toolkit (DMTK) openly available to the developer community. Researchers at the Microsoft Asia lab have released the toolkit on GitHub under an MIT (Massachusetts Institute of Technology) license, to encourage the use of multiple computers in parallel to solve complex problems. Its design builds on a parameter server-based programming framework, which allows big data machine learning tasks to be easily scaled, and flexibly and efficiently executed. The toolkit also contains two distributed machine learning algorithms, which can be used to train the world's fastest and largest topic model, as well as the largest word-embedding model.
This is a welcome move, especially after Google did something broadly similar.
This is a welcome move, especially after Google did something broadly similar.
A really good move to restore some faith in the programming community. Machine learning is a field that is much more complicated than desktop programming or user gui software design and will drive the very future of computers. The more programmers that can be stimulated to do research in the field of machine learning the better IMO. Far better than spawning nothing but hyped up "web programmers" which is what Microsoft was all about a few years back with all the .Net/mono crap they push on everyone.
This message was not sent from an iPhone because Peter Sellers really was a deviated prevert without a dime for the call
Now that the second and third largest tech companies have open sourced their machine learning algorithms for research use, it's time for the world's most valuable tech company to do the same. OK, Apple, show us your stuff!
--- Often in error; never in doubt!
People have wanted to talk to their computers and get the computers to do stuff for them for decades. Now that computers actually can listen and do stuff for you, people realize that to be useful, they have to be listening ALL THE TIME. Apparently this is not what people want anymore, so machine learning isn't as useful anymore.
AirBnB also released Aerosolve on Github.
That isn't 100% true. A small embedded chip that uses a Matched Filter/ANN to activate/deactivate a microphone would prevent the need to have "always on" microphone functionality beaming a continuous audio stream to the cloud.
https://github.com/Microsoft/D...
Does it go on forever?
I know we chastise the editors a lot for not specifying the meaning of an abbreviation, but I feel like in this case it is actually not helpful at all - it doesn't much matter what MIT stands for - it matters more what the MIT license is.