Your gait is your tell

Researchers have developed a new way to identify you: they can tell who you are by the way you walk. The tech evaluates 24 different factors in your gait and is surprisingly accurate—it has a 99 percent accuracy rating. And if you’re trying to fake the way someone else walks, it won’t work. The system can root you out by evaluating the exact force you exert on the mat, along with other hard-to-copy factors.

While some people are concerned about this tool’s potential privacy violations, it looks like this system would be hard to scale. It can obviously identify only people it already has in its database, and to be in that database, you have to have walked on a special floor mat embedded with sensors for the initial recording, and then again for later identification. We’ll see if and how this is used in the future.

Don’t touch my child

In some cultures, it’s not appropriate for humanitarian aid workers to use traditional methods to discover if children are malnourished. Now, there’s a new method being tested that may be able to identify a child’s level of malnutrition from a photo. We’ll find out how successful it is at the end of July 2018, but the hope is it’ll be easier to assess nutritional needs in high risk areas and that information would help better distribute resources and funding.

Fooling facial recognition

With China exporting its facial recognition technology to Zimbabwe, effectively increasing the size of its already massive database of faces, and with US policing projects similarly harnessing facial recognition technology, the question becomes, what can I do to protect my privacy?

People are working on privacy technology, and one team of researchers from the University of Toronto seems to have made a breakthrough—at least with static photos. By using deep learning to make minor, almost imperceptible changes to pictures, they’ve developed a system that can (currently) fool facial recognition technology. After they present their findings at IEEE this summer, they hope to make their privacy photo filter available to the public.


More stories

Machine learning tool beats 58 dermatologists in predicting skin cancer
Luggage uses AI to follow you so you can walk hands free
AI uses Twitter data to predict violence levels at protests
Foreign schools uses AI to grade papers and monitor students’ emotions
Researchers are working to develop a tool that identifies bias in algorithms
Web app Imaginary Soundscape matches your favorite paintings to music
AI converts one style of music to another


If you’re interested in drones and AI (and maybe whales), listen to our latest episode by subscribing to Data Crunch on your favorite app (Apple PodcastsGoogle PlayStitcher, or Overcast).

If you want an interesting read about the potential future of AI, we recommend Life 3.0 by Professor Max “Mad Max” Tegmark.

Are you building a B2B product involving IoT, ML, or AI? If you need deep research help to drive emerging tech research, product strategy, and tech marketing that resonates with your customers, email and let us know what you’re working on.