The official start of summer is just days away, but no one is slacking around here. This week, we saw how Microsoft researchers used artificial intelligence (AI) to master what was thought to be the un-masterable “Ms. Pac-Man,” an addictive 1980s video game. We met the new Xbox One X, a console system designed to play the best games of the past, present and future. And we hung out at Microsoft’s Data Insights Summit in Seattle, where Power BI Premium was introduced. Let’s go:
It was a mix of nostalgia and sci fi, but in this case there was nothing “fi” about it: Microsoft researchers created an AI-based system that learned how to get the maximum score on “Ms. Pac-Man.” The researchers used a divide-and-conquer method that could have broad implications for teaching AI agents to do complex tasks that augment human capabilities. The team from Maluuba, a Canadian deep learning startup acquired by Microsoft earlier this year, used a branch of AI called reinforcement learning to play the Atari 2600 version of “Ms. Pac-Man” perfectly. Using the divide-and-conquer method, the team achieved the maximum score possible: 999,990.
image: https://mscorpmedia.azureedge.net/mscorpmedia/2017/06/DCU_featured-image_637x478.jpeg
https://mscorpmedia.azureedge.net/mscorpmedia/2017/06/DCU_featured-image_637x478-300x225.jpeg 300w, https://mscorpmedia.azureedge.net/mscorpmedia/2017/06/DCU_featured-image_637x478-75x55.jpeg 75w" sizes="(max-width: 637px) 100vw, 637px" style="display: inline-block; margin: 7.75px auto 0px; width: 647px;">
Between May 2014 and October 2016, Microsoft’s U.S. operations received 118,417 complaints about tech support scams in which a payment was not made, and 19,878 in which victims made payments to the scammers. In this Power BI data visualization, the blue lines represent complaints without payment and the yellow lines represent complaints with payments.