10 Bits: the Data News Hotlist
This week’s list of data news highlights covers April 23-29, 2016 and includes a Microsoft initiative to store data on synthetic DNA and a new open-source artificial intelligence training program.
The UK House of Commons Science and Technology Committee has approved the formation of a Council of Data Ethics to address ethical issues related to data science, including privacy, anonymization, and sharing data for public benefit, conduct technical research, and help address the shortage of digital skills in the United Kingdom. The Council of Data Ethics will operate as part of the Alan Turing Institute, the UK government’s national institute for data sciences.
Twitter’s artificial intelligence team has developed an algorithm that can interpret the contents of live video feeds on its Periscope app in real time. Because video data is challenging to analyze, Periscope and other video services typically recommend content to users based on the habits of similar users, but this approach is not very effective for live content. Twitter’s new algorithm provides more sophisticated curation of live content, and it could also be useful for more effectively filtering out content that violates Twitter’s terms of service.
Montreal has published its first police data set on the city’s open data portal as part of Montreal’s Smart City Initiative, which includes a focus on government transparency. The data set includes 10,860 breaking-and-entering offences for 2015 and the beginning of 2016, with the locations of each crime approximated to the nearest intersection to avoid exposing sensitive information about victims and their homes. City officials will also publish visualization tools to allow members of the public to easily analyze the data.
Microsoft has partnered with genomics company Twist Bioscience to experiment with using synthetic strands of DNA to store data, after initial testing has shown it could be possible to write and recover data from strings of molecules. The small size and complexity of DNA strands means that, if successful, this technique would allow Microsoft to store approximately one zettabyte (one trillion gigabytes) of data in one gram of DNA.
Pennsylvania’s Department of Environmental Protection will deploy connected particulate matter sensors in eight counties with fracking sites to monitor the impact fracking has on air quality. Fracking can release large amounts of pollutants into the air, and the eight counties selected are in rural regions that environmental advocates say have gone under-monitored, limiting the government’s ability to understand and manage potential health risks from fracking in these areas. The agency will make the data freely available online.
Researchers at the University of Wisconsin-Madison have developed a system for drones that can detect the presence of specific materials found in explosives or nuclear devices just by flying over the area. The technology, similar to that used in security checkpoints at airports, covers an area in minor, harmless levels of radiation, and sensors on the drone then monitor for gamma rays or other types of particles, such as those emitted by explosives. The technology is still new and requires additional testing, but the researchers believe it could have a wide variety of valuable applications, including identifying unexploded landmines in former war zones, locating people trapped under rubble after a natural disaster, and mapping mineral deposits.
OpenAI, the artificial intelligence research nonprofit launched in December 2015 with backing from Elon Musk, has released OpenAI Gym, an open source tool for developing and testing machine learning algorithms. OpenAI Gym includes code and templates designed to make it easy for users to develop their own algorithms, train them through trial and error, and share their results with others.
Social media app Foursquare, which is based around users sharing geolocation data to “check-in” at different establishments, predicted the decline in sales at Chipotle restaurants weeks ahead of time and seemingly more accurately than Wall Street. Based on analysis of the declining number of check-ins at Chipotle restaurants, Foursquare’s chief executive officer Jeff Glueck estimated on April 12 that Chipotle’s first-quarter sales would fall by approximately 30 percent, and on April 27, Chipotle reported a 29.7 percent decline in sales.
The European Union’s new Sentinel-1b Earth-observation satellite has transmitted its first radar imagery back to Earth after its launch on Monday, April 25, indicating the launch was a success and the satellite is functional. Sentinel-1b will undergo a series of testing to ensure it is completely operational by the time it reaches its intended orbit in September, when it will work with the Sentinel 1-a satellite, launched in 2014, to map the surface of the Earth every six days.
Researchers at Waseda University in Tokyo have developed an artificial intelligence program that can realistically turn black-and-white photographs into color ones without any human intervention. Similar programs exist, but require human input to guide the colorization process, such as providing the program with a similar, color version of a photograph as a reference. The program relies on a machine learning technique called convolutional neural networks that mimic the visual cortex of a cat’s brain for analyzing images.
Image: European Space Agency.