Microsoft Corporation (NASDAQ:MSFT) has open sourced a toolkit that uses deep neural networks as part of an effort that seeks to speed up the development of its artificial intelligence capabilities. Computational Networks Toolkit (CNTK) is the new toolkit that the software giant says will speed up AI and speech recognition projects.
Open Sourcing AI Tool Kits
Alphabet Inc. (NASDAQ:GOOGL) just like Microsoft Corporation (NASDAQ:MSFT) has already released its deep learning software that is already available to developers for free. The TensorFlow machine learning system is the search giant’s toolkit that currently powers its photos in search functionality.
Researchers first made CNTK available last April but under an academic license. The fact that it is now freely available on Github should make it more useful to deep learning startups that are looking to process a good amount of data in real time.
While the previous license made CNTK accessible to academics, its usage was limited. Open sourcing it now ensures its usage can be expanded beyond the academic circles and more into the production world. The fact that it is highly optimized for speed should make it perfect for handling huge chunks of data.
Microsoft Corporation (NASDAQ:MSFT) is already ramping up support for its toolkit saying it is more efficient than anything that is already available. The fact that the CNTK toolkit can leverage the power of graphics processing units could give it an edge with developers. The toolkit according to the software giant is designed to scale beyond a single machine while taking advantage of GPUs on multiple machines for enhanced performance.
Redmond-based Microsoft Corporation (NASDAQ:MSFT) has already paired CNTK with its own GPU system and in the process trained its deep neural networks for speech recognition on Cortana. Thanks to the integration, the virtual assistant is now ten times faster than before.
CNTK is not the only deep learning arsenal developed by Microsoft Corporation (NASDAQ:MSFT) in the recent past. Last year the software giant quietly launched distributed machine learning toolkit DMTK that it says is well equipped for handling large amounts of data.