Remember How-old.net and #howoldrobot? The site where everyone was uploading pictures of themselves and Justin Bieber for age-checking? Miccrosoft’s corporate vice president of Information Management & Machine Learning, Joseph Sirosh, has taken to the company’s Machine Learning Blog today to share about the magic behind one of the “coolest” viral Microsoft project to come out in a while, and it has all to do with Cortana.
To be specific, the project is powered by components of the Cortana Analytics Suites, a collection of perceptual intelligence APIs for big data and analytics first unveiled in July this year. Started as a fun demo of the suite’s power for Sirosh’s //Build keynote, How-old.net quickly went viral and in 7 days had had 50 millions people uploading half a billion images to test the algorithms.
In his post, Sirosh reveals some interesting facts about what his team learnt through this data, including some tips to look younger. The team also learned important lessons about building cloud-based applications that can scale with data, and the advantage of a mobile-first, cloud-first project in rapid experimenting with minimal development effort.
The blog post also mentions a newly-added community feature for How-old.net: results now also include guesses from online users (anonymous of course.)
Overall, How-old.net has been a fun and powerful way of showcasing the power of Cortana Analytics and Microsoft’s machine learning prowress. Their next focus will be to simplify the process, in line with the recent commonly-heard mantra of customer empowerment. In the words of Sirosh:
We believe that analytics should be agile, simple and beautiful – and accessible to everyone. I even like to say that performing analytics should be as easy as following a simple cooking recipe.
Further reading: analytics, Big Data, Cortana Analytics Suite, Machine Learning, Microsoft