11 stories
today

Microsoft pioneers new ‘machine teaching’ technology to bring machine learning to the masses

Microsoft pioneers new 'machine teaching' technology to bring machine learning to the masses

Tech companies are constantly building and testing technology which could cause the next paradigm shift in how the world communicates, creates, and consumes. Many big names including Google, IBM and Microsoft are investing in machine intelligence and machine learning. Now Microsoft believes they have created the next generation of machine learning which they call machine teaching. While the name ‘machine teaching’ does not instantly communicate the purpose or intent of the new tech the underlying concept is simple.

Essentially, like Henry Ford brought the automobile to the masses, Microsoft wants to bring machine learning to everyone. Many companies are focused on making their machine learning algorithms more accurate, but Patrice Simard believes more advances can be driven by bringing machine learning to the masses. Patrice Simard, a distinguished engineer who is leading a new machine teaching research project at Microsoft Research, plans to focus on how to make the tools and UI possible for non-experts to create helpful and valuable machine learning capable systems.

Machine teaching could be used by doctors to sift through medical records for patters, or a restaurant owner could use data they have to train machine intelligence to predict demand and prepare their staff and inventory. At its core machine teaching doesn’t aim revolutionize machine learning by making predictions better or faster, but by exposing machine learning practices to more systems. Microsoft believes the next tech revolution will occur outside the public eye in machine intelligence.

Developers can take advantage of machine teaching today by implementing code which preforms face recognition, or translation tools. The Skype Translator was developed and works using machine learning, yet bringing real time translation to Skype was done by reusing complicated natural language understanding and language translation work by other teams. The same principle could be applied to other apps who’s developers are not experts but still want to leverage sophisticated technology. Ideally our mobile apps will become smarter and more useful without requiring huge teams of multidisciplinary developers.

Further reading: , , ,