The team behind the Whetlab startup wants to develop A.I.-style technologies that would make it easier for companies to implement a machine learning system. Its website says that the company's patent-pending system is designed to architect and tune a machine learning system in an in-house setting as opposed to the traditional way of hiring experts to do similar work. It even claimed that the system outperformed today's human experts when it comes to system configuration which is needed to solve even the most difficult issues.
Developed by a research team of five members from Toronto, Harvard and Sherbrooke universities, the technology has been designed to be applied on things such as computational biology, speech processing and visual object recognition. The startup was only about a year old when the acquisition took place.
While Twitter chooses to remain silent on the other details of the acquisition such as the financial information and reasons for acquiring, a number of speculations have actually been made.
First among them deals with the acquisition timing as the decision was made during a tough time for Twitter when CEO Dick Costolo confirmed that he is leaving his post. Perhaps the company is looking for new opportunities and is coping up with missed revenue and user number forecasts.
The move is described by David Schubmehl, a research director at IDC, as a "general purpose acquisition that could give Twitter a number of options, including around targeting of ads and positioning of relevant tweets based on interest. As far as I know, this is their first foray into AI or machine learning."
Twitter's interest in machine learning technology can bring a number of advantages for the company. Using it would allow a better way of identifying users' posts which in turn provides a deeper understanding of its user base; it could enhance the way it detects spam posts; it could improve the company's ad targeting ventures; it could provide better means on taking advantage of Twitter's huge storage of data; it could promote various opportunities for recommendations, and ideas to get things more personalized, and so on.
"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," says Whetlab on its announcement page.
Whetlab advised that it is shutting down its closed beta on July 15, 2015. Current beta users can download the data from their past experiments through the site as raw JSON. They can also use the company's command-line utility which can come in either JSON or tab-separated format. Lastly, it recommends users to try SMAC, Spearmint, or Hyperopt as alternative systems for machine learning optimization.
