The Stack Archive

Microsoft moves its deep learning CNTK toolkit to GitHub

Mon 25 Jan 2016

Xuedong Huang Microsoft CNTK

Microsoft has today announced that it will be moving its machine learning Computational Network Toolkit (CNTK) from its own hosting site, CodePlex, to GitHub and the MIT open source license.

The move marks an effort to make it easier for developers to collaborate on building their own deep learning applications using the CNTK. Under the CodePlex license, access was restricted to academics only and it was wholly targeted to this audience. Now opening the project to GitHub, Microsoft hopes to attract a greater number, and a wider variety, of developers.

Microsoft chief speech scientist, Xuedong Huang, underlined in today’s announcement that its CNTK is highly optimised for speed – “The CNTK toolkit is just insanely more efficient than anything we have ever seen.” He refers here to projects under way at Google (TensorFlow) and elsewhere, including Torch, Theano and Caffe.

Huang explained that CNTK is a key component in Microsoft’s personal assistant, Cortana, where it is used for speech recognition. He added that the software is also useful for other applications, such as image recognition, and natural language processing. For example, Huang noted that CNTK powers several projects which rank web search results and predict which ads people will click on.

According to Microsoft, CNTK is unique in its ability to run on a single core, as well as on a large cluster of GPU-based machines. It also claims that the tool-kit can scale across more machines than any other rival project.

The framework is written in C++ and is integrated with the Nvidia cuDNN 4 library. It offers image, speech, and text demonstrations. “We do not support Python yet,” said Huang. “We are very anxious to really get feedback from the community and improve the platform.”

Last year, the U.S. giant also released its Distributed Machine Learning Toolkit (DMTK), albeit a quieter launch. The project is focused on improving the efficacy and flexibility of analysing big data.


AI developers Microsoft news open source
Send us a correction about this article Send us a news tip