This week’s list of data news highlights covers April 22-28, 2017, and includes articles about a new self-driving car pilot in Phoenix and an AI company building a dictionary for dolphins.
Sens. Cory Gardner (R-CO) and Gary Peters (D-MI) have introduced the Preserving Data in Government Act to prohibit federal agencies from removing access to datasets they have already published as open data. The Federal Records Act already required federal agencies to preserve their data, however there is no requirement for agencies to maintain this data in open and machine readable formats. The bill would prevent agencies from removing access to open data after it has been online for 90 days, as well as limiting the data’s accessibility, such as by reposting it in non-machine-readable formats.
The National Football League (NFL) Players Association has partnered with wearable device company Whoop to allow NFL players to capture their biometric data to share and analyze it as they please. Whoop’s wrist-worn devices collect players’ biometric data 100 times per second during games and training and transmits this information to a server for analysis. The NFL Players Association will analyze the data to study how to improve player safety, but players are free to use this data for personal purposes as well as sell access to their data, such as to a sports network.
The University of Washington’s Center for Game Science and the nonprofit Allen Institute for Brain Science have developed a video game called Mozak that allows researchers to crowdsource the development of 3D models for neurons in the brain. Due to the complex shapes and sheer quantity of neurons, it can be very challenging for researchers to build useful models of the human brain. Mozak, which averages 200 players a day, uses gamification techniques such as awarding points and public leaderboards for high-scorers, to encourage players to trace the structure of neurons in microscope imagery so software can convert this data into 3D models.
Google has announced new features to its search algorithms designed to reduce the visibility of extremist and purposefully misleading content. Google now has a team of 10,000 people focusing on evaluating the quality of sources for pages that place highly in search results to develop guidelines for page ranking that demotes low-quality content. Google has also adjusted its algorithms to increase the likelihood that authoritative sources rank higher than untrustworthy sites. Additionally, Google has implemented direct feedback tools for users to flag low-quality content in search results so it can use this information to further improve how its algorithms distinguishes between high- and low-quality content.
Alphabet’s self-driving car company Waymo has launched a pilot test in Phoenix, Arizona that will allow select residents to freely use its autonomous cars as they go about their daily lives. Phoenix residents can apply to participate in the program, and those accepted will be able to summon Waymo cars—with a human driver ready to take the wheel in an emergency—with a smartphone app and use them for free transportation. Though Waymo has tested its cars with thousands of rides internally, offering them for public use allows it to gather more useful, real-world data about how people use self-driving cars.
Amazon has launched a new smart home device called the Echo Look that has a camera and allows users to evaluate their outfits with machine learning. Users can have Amazon’s AI assistant Alexa take a photo of themselves with the Echo Look and use an application called Style Check that analyzes the photo to compare different outfits and generate a style rating. Based on a user’s wardrobe, Style Check can also recommend new articles of clothing to buy from Amazon.
Swedish language analysis startup Gavagai AB and researchers from Sweden’s KTH Royal Institute of Technology are using AI to monitor bottlenose dolphins and build a dictionary of the noises they use to communicate. The researchers will record captive dolphins at a wildlife park and use natural language processing software to analyze the distinct sounds dolphins make to each other and attempt to discern their meaning.
Norwegian telecommunications firm Telenor is deploying a low-power wide-area (LPWA) wireless network in several cities in Norway that it will provide free access to for Internet of Things startups and students. Beginning May 1, 2017 participating startups in Trondheim, Tromø, and Oslo will be able to experiment with IoT applications over the network and will have free access to development kits and Telenor’s backend cloud portal.
Researchers at Nottingham University have developed a series of machine learning algorithms capable of predicting a person’s risk of cardiovascular disease more accurately than humans. The researchers developed four different types of algorithms and had them analyze a dataset of nearly 400,000 patients in the United Kingdom. Each algorithm’s approach proved better at predicting the likelihood a person would develop cardiovascular disease than humans, with the best-performing algorithm—a neural network—correctly identifying 7.6 percent more patients that would go on to develop cardiovascular disease.
Stanford University biology professor Manu Prakash has launched a citizen science platform called Abuzz that enables users to upload recordings of mosquito buzzes from their smartphones, which can be used to map the locations of mosquito species that carry malaria. Though there are thousands of species of mosquito, only several dozen are capable of transmitting malaria. Abuzz will analyze these geo-tagged audio recordings to identify mosquito species based on the frequency of their wing beats to help guide efforts to eradicate mosquito populations that could be carrying malaria.
Image: Hamid Elbaz.