Microsoft released its Computational Network Toolkit (CNTK) on GitHub, making the tools that its own researchers use to speed up advances in artificial intelligence available to a broader group of developers.
Xuedong Huang, Microsoft’s chief speech scientist, and his team were anxious to make faster improvements to how well computers can understand speech, and the tools they had to work with were slowing them down. The group of scientists set out to solve this problem, using a homegrown solution that stressed performance over all else.
Their hard work paid off — in internal tests, CNTK has proved more efficient than four other popular computational toolkits that developers use to create deep learning models for things like speech and image recognition, because it has better communication capabilities.
“CNTK provides algorithms to carry out both forward computation and gradient calculation. Most popular computation node types are predefined and users can easily extend node types under the open source license,” wrote Huang. “With the combination of CNTK and Microsoft’s upcoming Azure GPU Lab, we have a modern, distributed GPU platform that the community can utilize to advance AI research.”
For more on CNTK and its advantages for deep learning projects, large and small, check out the Microsoft Next blog.