This week, how athletes at the Winter Olympics are staying warm and how drones were used in the Opening Ceremonies. Plus: the racist nature of facial recognition technology and what to watch out for if you’ve got an Amazon Echo in your home.
Battery-powered ski jackets and pants keep athletes warm in PyeongChang
The temperatures in South Korea haven’t been too apocalyptic. On average, it’s been between -10 and -15 Celsius, For many Canadians, that’s only moderately cold.
But if you’re spending your entire day in that weather, and you’re competing against the best athletes in the world, you want to stay warm.
To that end, Canada’s downhill ski team has battery-powered pants to keep warm in between runs.
And the American team wore self-heating jackets during the opening ceremonies. The battery-powered apparel was designed by Polo Ralph Lauren, and were available to the general public for the reasonable price of US$2,500. They are sold out.
At least one athlete has been using a Lungplus which is a heat recovery device similar to those you’ll find in energy-efficient homes. This is intended to minimize the heat lost during breathing.
There hasn’t been any research to prove whether using the Lungplus can result in an improved performance. But if an athlete is more comfortable being in the cold, there could be at least a psychological effect.
Your computer is racist. Who’s at fault?
Joy Buolamwini is working on a PhD at MIT’s Media Lab. An African-American, as an undergraduate in computer science she had the experience of facial recognition software not working with her. The systems didn’t even recognize her face as a face, she told the New York Times.
In new research, the Rhodes scholar working under a Fullbright scholarship, has demonstrated that facial recognition technologies in use around the world are racially biased.
She had various programmes look at 1270 “unique faces” and make a determination of gender. White males were correctly identified 99% of the time. Darker skinned females were correctly identified 65% of the time.
Even guessing, there’s a 50% of getting the answer right.
Part of the problem is the programmers. Artificial intelligence learns as it crunches data and makes errors. It turns out that many data sets being used to “teach” computers facial recognition use photos of white (80%) males (75%).
Buolamwini’s paper is published in Proceedings of Machine Learning Research.
Ad agencies are salivating at the possibilities after Alexa tries to place an order for cat food
Drones steal the show at opening ceremony of 2018 Winter Olympics
If you watched the opening ceremonies last week, you saw an impressive aerial presentation that included images of a snowboarder and the Olympic rings.
Those images were all created by LED-equipped drones. More than 1200 of them, each about a foot long and weighing 8 ounces.
The performance was courtesy of Intel and its Shooting Star division.
What we saw on TV was recorded earlier. There were plans for a live performance, but they were scrapped. No official reason was given, but it was likely due to the cold, windy weather.
There are regular live performances, with a smaller number of drones, that are occurring during medal ceremonies.