Microsoft Puts Deep Learning Toolkit On Github

Updated on

Tech giant Microsoft has decided to open-source its deep learning toolkit, in a move which makes the company stand out from bitter rival Google.

On Monday Microsoft announced that its internal deep learning toolkit would be released on GitHub, marking further progress in the company’s open-source sharing initiative.

Microsoft releases efficient deep learning toolkit on GitHub

The Computational Network Toolkit (CNTK) is a unified deep learning toolkit that defines neural networks as a series of computational steps using a directed graph. According to Microsoft the system has proven far more efficient than competing tools such as Google’s TensorFlow.

So far the company has used CNTK internally in its work on speech recognition products such as Cortana. However executives maintain that it could be useful to a huge number of users, not just those focused on deep  learning. Another possible target audience are those companies that need to process large amounts of data in real time.

Academics were granted access to the software last year via Microsoft’s CodePlex site, albeit under a more restrictive custom license. Now that it is on GitHub the software is available under the less restricted MIT license.

Open-source sharing journey takes another step

In November 2015 Microsoft made a similar move with its Distributed Machine Learning Toolkit (DMTK). Google also then released its TensorFlow system under an Apache 2.0 license.

So far Microsoft engineers have been controlling DNTK using GPU-based computers, but they maintain that such power isn’t required to get the most out of the system.

“The combination of CNTK and Azure GPU Lab allows us to build and train deep neural nets for Cortana speech recognition up to 10 times faster than our previous deep-learning system,” explained Xuedong Huang, Microsoft’s chief speech scientist, in a December blog post. “We’ve seen firsthand the kind of performance CNTK can deliver, and we think it could make an even greater impact within the broader machine learning and A.I. community.”

Watch out for another wave of innovation in the fields of machine learning and artificial intelligence now that access to the tool has been made easier for ordinary developers.

Leave a Comment