U.S. – Microsoft Inc. (NASDAQ: MSFT) released an updated version of Microsoft Cognitive Toolkit, a machine learning system that can speed up research in areas such as image and voice recognition and search relevant on Nvidia GPUs and CPUs.
The latest version is available on GitHub under an open source license. Developers can now work with the toolkit with Python or C++ programming languages. The update allows researchers to code a type of artificial intelligence called reinforcement learning.
Reinforcement learning is a branch of machine learning that enables computers and software to automatically determine the ideal behavior or answer for a particular context, to optimize its performance.
What is Deep Learning?
A group of computer scientists from Microsoft developed the toolkit as a means to do their research in a quick and efficient way. The system moved beyond voice recognition and transformed into the core platform of a group of flagship products like Cortana.
Nvidia collaborated in the growth of the toolkit, and it now brings advances to the broad learning field. Likely, developers optimized the software to take advantage of Nvidia hardware and Azure networking capabilities.
The tech overlord hopes the tool will help both startups and major technology companies develop deep learning technologies such as speech understanding and image recognition.
Deep learning is an artificial intelligence technique that requires computers to recognize patterns and inputs. Programmers must give the machines large amounts of data –training sets – to complete a task.
For example, an instruction set could show the computer pictures of fruits, so the it learns to recognize images of fruits on its own. The more data, the better, so every time the code finds a new rarity, it can refine the algorithm and increment its accuracy.
Microsoft is seeing the results of its machine learning research
The company says the updated Microsoft Cognitive Learning is better and faster than ever. The tech giant also assures the platform will work particularly well on big databases with several computers.
The toolkit aims to deploy into large-scale business as it wants to deliver deep learning across multiple GPUs to develop consumer electronics and professional offerings.
It is also essential to speed up breakthrough advances. Last week, Microsoft Artificial Intelligence and Research Branch announced they had created their first technology that recognizes words in a conversation as well as a person.
The team says the machine learning toolkit significantly improved the speed of their research.
Recently, manufacturing enterprise Liebherr Group and Microsoft teamed up to build a smart refrigerator for pharmacies. It will offer remote monitoring and predictive maintenance, and it will work on the tech giant’s machine learning platform.
Source: Microsoft Blog