Redmond, Washington – Microsoft Corporation (NASDAQ: MSFT) announced it has uploaded its Computational Network Toolkit (CNTK) from CodePlex to the acclaimed open-source host GitHub. The move will make the toolkit, which the company claims is “more efficient” than other solutions used to create deep learning models, more accessible to developers.
CNTK was launched in CodePlex last April with restrictions by an academic license, meaning that the open-source deep-learning toolkit was virtually unused beyond scholarly use. Now that it is uploaded on GitHub, more developers will be able to join the mix.
“It is our hope that the community will take advantage of CNTK to share ideas more quickly through the exchange of open source working code,” Microsoft’s upload on GitHub states.
Microsoft’s toolkit outperformed other computational solutions used to make deep-learning models for image and speech recognition, among other applications. Xuedong Huang, Chief Scientist of Speech Research and Development, claimed that CNTK was “insanely more efficient” that any other toolkit of its kind, with more capable communication capacities that have been added to it.
CNTK was compared in the internal tests with Torch 7, Theano and Caffe, as well as Google’s TensorFlow that was recently open-sourced. One of the advantages CNTK has over others is that it has the ability to run on a single core or on a “large cluster of GPU-based computers,” as described by Microsoft Principal Development Manager Chris Basoglu.
The toolkit, which supports both GPU and CPU, was developed after Huang and his team realized the tools they used were slowing them down as they were trying to improve the way computers understood speech. They decided to build their own and are now using it on a set of computers that use graphics processing units (GPUs).
Moreover, the research team found that GPU is highly efficient when it comes to processing algorithms used in technology that can recognize and comprehend not only images and speech, but also movements. Cortana, Microsoft’s well-known virtual assistant, uses the capabilities of CNTK for speech recognition.
The company is interested in inviting users to take advantage of its CNTK by allowing them to have access to the same resources it uses for internal needs. This is a way the Microsoft team found to contribute to the progress of advances in artificial intelligence.
Source: Tech Times