This week’s list of data news highlights covers October 31 – November 6, 2015 and includes articles about how Facebook is teaching an artificial intelligence program to play Go and how Australia’s government is supporting startups that use open data.
The U.S. National Science Foundation (NSF) has established four regional hubs across the country to bring together academic, industry, and government stakeholders to advance the study of big data and conduct research into how data can deliver social and economic benefits in areas such as precision medicine, smart cities, and public safety. NSF will also award up to $10 million to support collaborative efforts that work with the hubs to advance three specific issues related to big data: improving progress toward the Obama administration’s grand challenges—ambitious initiatives to solve pressing national problems with technology; automating the big data lifecycle; and increasing the accessibility and usability of valuable government data sets.
The U.S. Census Bureau has announced that the Census Enterprise Data Collection Processing initiative (CEDCaP), the agency’s plan to modernize how it collects and disseminates data, will reduce the cost of the 2020 census by $5.2 billion. CEDCaP will focus on consolidating the over 100 disparate data management systems used in the 2010 census into a single platform, and the agency has partnered with AT&T to equip its 20,000 field workers with technology that can more cost-effectively collect and report survey data in real time.
Researchers at the University of Rochester have developed a system to analyze Instagram photos to reveal insights into the drinking patterns of underage teenagers. The researchers developed an algorithm that flags Instagram posts that contain words associated with drinking, such as “wasted,” and “shots,” and used facial recognition software to determine the age, race, and gender of the subject of flagged posts. The researchers hope their analysis could help guide parents’ efforts to curb underage drinking.
Facebook has announced that it is attempting to develop artificial intelligence technology capable of beating humans at the strategy board game Go. Though computer systems have already successfully beaten the most skilled human opponents at chess, Go relies on advanced visual pattern recognition that humans are adept at, but which programmers have struggled to instill in software. By developing an artificial intelligence system capable of winning at Go, Facebook hopes to improve how its artificial intelligence technology can accomplish other tasks that require advanced pattern recognition, such as voice recognition tools for its users with visual impairments.
The Australian government has partnered with startup incubator Pollenizer to create DataStart, a pilot project to support startups that rely on open data. DataStart will select a startup that uses data from Australia’s open data portal in January 2016 and provide it with technical and business development support, as well as assist it with accessing government data, for an incubation period of nine months.
Google has released a new feature for its smartphone email Inbox app called Smart Reply, which can interpret emails and generate suggested responses. Smart Reply relies on a machine learning system powered by artificial neural networks—algorithms designed to mimic how the human brain processes information—to generate appropriate responses to emails based on their content and context. Users can send their choice of three response options with just one tap of their screen if they do not wish to take the time to write out a response themselves.
Canada’s Ministry of Innovation, Science, and Economic Development has announced that it would reinstate the country’s mandatory long-form census, which was abolished by the previous administration in 2010. The new census will replace the less comprehensive, voluntary census distributed in 2011 that many criticized as insufficient for collecting data valuable to the public and private sectors alike. Canada will conduct its long-form census in 2016.
The United Kingdom’s Cabinet Office has announced a series of initiatives to improve the UK’s open data policies and increase the utility of open data. The government will work on developing application programming interfaces to make it easier for agencies to use open data and will establish a program to train government workers to use data more effectively. Additionally, the government will establish the Data Leaders Network, which will have a steering group of private sector and civil society leaders, to evaluate open data policies.
The U.S. Department of Homeland has announced plans to equip personnel of the Coast Guard, Customs and Border Protection, and Transportation Security Administration with a wearable device called the Human Portable Tripwire (HPT) that can detect potential nuclear threats. HPT’s sensors constantly monitor the environment for signs of radioactive material and can alert the wearer about the source of the radiation to guide their response.
The Maine Health Data Organization (MHDO), an independent state agency focusing on increasing healthcare cost transparency, has launched a new website called CompareMaine to allow Maine residents to compare the costs of hospitals and factors that could indicate the quality of healthcare providers, such as patient satisfaction and healthcare rates. MHDO created the website to help residents make more informed decisions about where to seek treatment, as costs for similar treatments can vary widely from hospital to hospital.