Amazon and Microsoft Now Announce a Joint Project: Gluon

Posted by Augusto Alvarez on October 31, 2017
No ratings yet.

It may be a surprise for some, but when two cloud – or even better- technology titans are facing each other, there’s going to be some collaboration and “friendly” competition. In this case, Amazon and Microsoft just announced the release of new deep learning library for developers called Gluon.

Gluon Logo

Gluon was defined by these two companies as “an open source deep learning interface that allows developers to more easily and quickly build machine learning models without compromising training performance”. But, what exactly does this represent? It’s acting as an interface that lets developers, of different skill levels, use Python and pre-built deep learning templates to simplify the building of models to run neural networks.

Neural networks are one of the key aspects of machine learning, it is represented by a system or program that process the information that is presented based on the dynamic responses it receives. The deep learning concept represents the set of techniques for learning used in these neural networks.

Graphical example of a neural network

Graphical example of a neural network

The neural network definition is usually a manual process, which can take days from developers and architects. Gluon was created by Amazon and Microsoft as a way to generate simpler ways to build and facilitate the engines learning, so more and faster optimizations could be introduced in the processes.

In this process of looking for improvement, these two companies “got to talking and found we had a similar vision: to use these techniques to reduce the complexity of machine learning, making it accessible to more developers”.

Sheme - Make a prediction - Calculate the errors - Adjust the prediction

Gluon is available today in AWS, using Apache MXNet, and will be available soon in the Microsoft Cognitive Toolkit. Since this is an open source platform, it is expected that more and more frameworks will use Gluon over time.

Some of the key benefits and innovations Amazon announces around Gluon:

  • Friendly API: Gluon networks can be defined using a simple, clear, concise code. Easy for developers to learn and create.
  • Dynamic networks: Dynamic networks are easier to manage, and with Gluon, developers can easily ‘hybridize’ between these fast symbolic representations and the more friendly, dynamic ‘imperative’ definitions of the network and algorithms.
  • The algorithm can define the network: Instead of separate definitions, the algorithm can adjust the network dynamically during definition and training.
  • High performance operators for training: All the Gluon benefits can be introduced without slowing the training process. Gluon can efficiently blend together a concise API with the formal definition under the hood, without the developer having to know about the specific details or to accommodate the compiler optimizations manually.

Gluon API is available for download in GitHub.


Related materials:

Views All Time
Views Today
Return to all posts

How to Create a Shared Mailbox in Microsoft Exchange Server 2016 with PowerShell
Microsoft is Announcing Azure File Share Snapshots in Public Preview

Please rate this

Leave a Reply

You must be logged in to post a comment.