The Microsoft Cognitive Toolkit was originally launched in beta back in October 2016. It provides a set of robust intelligent services that enable developers to draw on AI within their own products. The revised version launched this week includes “hundreds” of new features that combine to make a simpler and more powerful toolkit.
Microsoft originally built the Cognitive Toolkit to aid the development of its own internal services. It was devised as a method of accelerating the previously costly training procedures required to launch new neural networks. The system was put to work preparing AI for Microsoft’s advanced speech recognition and search technologies.
As the Cognitive Toolkit grew to include more services, Microsoft realised that its contents could be useful outside of its own walls. The company has open-sourced the code to allow anyone to use it in their own products.
The features on offer have expanded to cover several other capabilities commonly powered by AI. Today, the Cognitive Toolkit comes with an impressive array of options that let developers intelligently understand speech, detect the contents of images and accurately identify human facial expressions. The machine learning algorithms that power the services are built to mimic the way in which the human brain thinks.
Microsoft detailed a few prominent examples of organisations that are already using the toolkit. One of them, Chesapeake Conservancy in Annapolis, Maryland, is using the system to create and train a neural network that will let it publish up-to-date land cover datasets with a resolution of one-metre. The new data is 900 times more detailed than the last-generation files.
The conservation team would take months to process all the data using traditional imaging methods. Microsoft’s AI cuts it down to a fraction of the time, allowing the organisation to rapidly identify areas of land that need to be protected. The conservation region being studied spans an area of around 64,000 square miles across six U.S. states.
“Originally, people handwrote their own mathematical functions and created their own neural networks with their own private code and figured out how to feed it with data all by themselves,” said Microsoft partner engineering manager Chris Basoglu. “But now the data is so large, the algorithms are so complex and optimization across multiple GPUs, CPUs and machines is so prohibitive that it is not feasible for someone to write their own. They need tools.”
While Microsoft is naturally talking most about these major national-scale uses of the Cognitive Toolkit, it’s also stressing that the technology is available to everyone and anyone. The company has designed the software to be accessible to people who have self-taught themselves programming through to those working on the frontiers of neural network development.
The code for the project can be downloaded from Microsoft’s GitHub page. Alternatively, developers can also access the resources via the cloud using the Cognitive Services APIs in Microsoft Azure.
