MICROSOFT IS SLOWLY PREPARING FOR FUTURE TRENDS

Over the years, Microsoft is not only interested in their familiar products like Windows, Windows Phone, Xbox,... but also invests in research and development of AI (artificial intelligence).

Specifically, over the past 20 years, Microsoft has focused on developing voice recognition technology such as Cortana virtual assistant, Skype Translator, and Oxford Speech APIs project. Microsoft researchers were the pioneers in the use of Deep Neural Networks (DNNs) for the operation of voice recognition. Earlier this year, Microsoft has introduced CNTK open source (Computational Network Toolkit) at ICASSP conference on April 2015.

CNTK is a united calculating network system describing DNNs as a series of calculations through a directed graph. Most popular computing nodes on the graph have been predetermined, and thanks to the open-source license, users can easily expand the note types to develop tools. With the combination of CNTK and upcoming Azure GPU Lab, Microsoft has a modern GPU platform that communities can use to promote AI research.

Since their CNTK launch on April, Microsoft has significantly improved memory efficiency of the engine with Azure GPU Lab. The combination of CNTK and GPU Azure allows us to build and customize DNNs, helping Cortana to identify the voices 10 times as fast as its previous systems. CNTK is also used in other parts of Microsoft to run several projects, such as classifying IMAGEnet and developing the model to distinguish advanced contextual structures. High performance that CNTK provides will affect the process of memorizing the data to your computer and the AI ​​community.

Before CNTK, Microsoft had put their faith in SnapLogic. SnapLogic noticed that they received $ 37.5 million from investors, including Microsoft, and this amount has saved them from the brink of bankruptcy. With this new funding stream, the capital of SnapLogic has risen about 90 million dollars and they can continue to improve the products in 2016. SnapLogic announced a partnership with Microsoft to support the analysis of Cortana, also like working with HCL Technologies, the fourth largest provider of India’s IT services to consult and implement services for applications and integrating data.

Agent in charge of Microsoft’s CNTK Project said his department could fully meet the needs of companies to develop CNTK, however they also wanted to share this resource with other researchers who were working on developing advanced memory capacity of AI. That is why Microsoft decided to provide CNTK as open source.

Obviously, this is a very smart move by Microsoft, they are holding the most powerful tool in the field, and the free provision to the community will make researchers not think much to their personal interests but devote and improve to help Microsoft. Surely, there are also many researchers only hope to use the best tools to so that they can freely serve their passions.

After announcing IT, it seems that things have gradually become clear, Microsoft does not assume funding and time to invest in a project that does not benefit anything. Apparently they hope SnapLogic with cloud computing will help them easily popularize CNTK wider to the community who will contribute to the development of Microsoft tools.