In another instance of how brilliant startups get absorbed by larger corporations in an attempt to enhance and improve their existing operations, machine learning start-up Whetlab has been acquired by Twitter !
For those who are unaware of the Cambridge based company, Whetlab has been working with researchers from Harvard, Toronto, and Sherbrooke universities over the past year to create a brand new technology that when implemented, will supposedly make machine learning easier for their internal systems. Well, considering the fact that most major companies rely on machine learning one way or another, the move to acquire Whetlab does not come as much of a surprise.
The system, as the company likes to put it, would not only offer an alternative to hiring experts to architect and tune a machine learning system but could also cut down the time taken in the process. Further more, if the claims made by the folks over at Whetlab are to be believed, the technology apparently managed to “outperformed the top machine learning researchers in configuring systems for the hardest cutting-edge problems.”
Speaking on the acquisition, Whetlab posted the following statement on its official website,
Over the past year, we have created a technology to make machine learning better and faster for companies, automatically. Twitter is the platform for open communication on the internet and we believe that Whetlab’s technology can have a great impact by accelerating Twitter’s internal machine learning efforts. This does mean, however, that we will be shutting down our closed beta on July 15, 2015.
Although the terms under which the deal was closed are not available at the moment, we do know that under the agreement, Twitter gets ‘exclusive access’ to virtually all of Whetlab, the technology as well as the five member team — including Ryan Adams, Hugo Larochelle, Jasper Snoek, Kevin Swersky and Alex Wiltschko — who will now work to implement this tech to enhance the machine learning efforts of the gigantic systems possessed by Twitter.
While we are only speculating, the technology could potentially be used in many scenarios by the micro-blogging service such as for spam detection, ad targeting, offering recommendations and personalization based on user preference and much more. New signups for the closed beta meanwhile have been disabled and the existing beta testers have been advised to download their data then migrate to other systems.