Microsoft has open sourced its own artificial intelligence (AI) deep learning tools. The Computational Network Toolkit (CNTK), used by Microsoft’s own researchers, is now available on GitHub, and is aimed at fuelling the speed of AI technology advancement. Here are the details.
The CNTK has gone through internal testing and has been proven to be more efficient than four other popular computational toolkits used for deep learning models to analyse speech and image recognition. This means the CNTK can reduce the time it takes to complete deep learning tasks used for AI research.
Interest around AI has skyrocketed thanks to the likes of Apple’s Siri and Microsoft’s Cortana bringing the technology to the mainstream. Late last year, Google open sourced TensorFlow, its second-generation machine learning system used to power the company’s image and speech recognition, foreign language translation offerings and a variety of other applications.
It seems Microsoft doesn’t want to be left behind by Google in the AI field. The CNTK puts Microsoft in a strong position to drive future AI research. The company itself is using CNTK on a set of powerful computers running a cluster of GPUs. But if you’re on a tight budget, the toolkit can be deployed on a single computer.
Microsoft researchers said CNTK will be useful to anybody from deep learning start-ups to more established companies that are processing a swathe of real-time data to fuel AI breakthroughs.
You can find the CNTK on Microsoft’s GitHub page.
[Via Microsoft Research Blog]