Four short links: 23 October 2018
Forecasting, Cars and Privacy, Quantum Communications, and Positive Communications
- Forecasting at Uber: An Introduction — Actually, classical and ML methods are not that different from each other, but distinguished by whether the models are more simple and interpretable or more complex and flexible. In practice, classical statistical algorithms tend to be much quicker and easier-to-use. An important message that isn’t getting as much airplay as the sales pitches: deep learning is unfairly good on some problems, but not all.
- The Next Data Minefield Is Your Car — GM captured minuted details such as station selection, volume level, and ZIP codes of vehicle owners, and then used the car’s built-in Wi-Fi signal to upload the data to its servers. The goal was to determine the relationship between what drivers listen to and what they buy and then turn around and sell the data to advertisers and radio operators. And it got really specific: GM tracked a driver listening to country music who stopped at a Tim Horton’s restaurant. (No data on that donut order, though.)
- Inside Europe’s Quest for an Unhackable Quantum Internet (MIT TR) — China has also built a land-based QKD communications network from Beijing to Shanghai that banks and other companies are using to transmit sensitive commercial data. China’s approach requires trusted quantum-classical-quantum repeaters every 10km, whereas the Dutch university at the center of this article is looking to use quantum teleportation. Interesting to see the Dutch are connecting universities, as ARPA did at the birth of the internet.
- GNU Kind Communications Guidelines — astonishingly useful set of specific and positive recommendations. No mention of consequences of violation.
