Published on January 15th, 2016 | by Joshua New0
10 Bits: the Data News Hotlist
This week’s list of data news highlights covers January 9-15, 2016 and includes articles about an artificial intelligence program that can recognize humor and a new algorithm designed to recognize the faces of endangered whales.
President Obama announced a new “moonshot” initiative to accelerate public and private efforts to cure cancer by increasing funding and other incentives for researchers and removing barriers to medical and research data sharing that inhibit progress. The initiative, headed by Vice President Biden, will solicit input from scientific and government leaders to identify research and investment opportunities and, beginning this month, will convene high-level federal leaders to improve how the government manages and supports cancer research.
U.S. Representative Darrell Issa (R-CA) has introduced a bill that would reform the Freedom of Information Act (FOIA), the 1966 federal freedom of information law that authorizes public access to government documents and information. The bill, called the FOIA Act, would create a single online portal for all public requests for government information, direct agencies to publicly report FOIA compliance, and require agencies to identify categories of data which they could publish proactively and streamline the FOIA process, which many have criticized as sluggish and unaccountable.
The U.S. National Center for Atmospheric Research (NCAR), a federally funded research and development center, has announced that it is upgrading its weather forecasting system with a supercomputer capable of 5.34 quadrillion calculations per second, which is more than twice as powerful as its current system. The new supercomputer, named Cheyenne, will allow researchers to develop more detailed models and simulations about weather patterns, which can be particularly valuable for preparing for severe weather events such as heavy flooding, wildfires, and solar storms. Cheyenne will be operational in 2017.
Researchers at the Virginia Polytechnic Institute and State University have developed a machine learning algorithm capable of recognizing humor in specially created animated scenes without any context of the scene. The researchers developed a system for creating animated scenes with interchangeable models for human characters, animals, and objects, and trained its algorithm on a database of 6,400 of these scenes tagged with scores indicating their humor levels. The algorithm was able to identify humor in new scenes based on factors such as abnormal or unexpected positioning of the models, and could even alter the scene to be less humorous with 95 percent reliability.
The U.S. Department of Transportation (DOT) has announced a series initiatives to accelerate the development of autonomous vehicles, such as increasing funding and reducing regulatory obstacles to integrating innovative automotive technologies. President Obama’s fiscal year 2017 budget proposal would allocate $4 billion over 10 years for autonomous vehicle pilot programs. This year, DOT will develop safety guidance and establish standards for testing, as well as work with public and private sector stakeholders to develop a model state policy for autonomous vehicles.
Fox Studios has partnered with bioanalytics software company Lightwave to study how audiences react to movies based on their biometric data. In fall 2015, test audiences for the new movie The Revenant wore biometric monitors on their wrists which recorded data about their heart rates, movements, and other factors that indicate excitement. Fox Studios analyzed how changes in this data, such as spikes in heart rate, related to specific scenes in the movie, revealing that The Revenant contained 14 “heart-pounding” moments. This approach could improve how studios and filmmakers produce movies to better stimulate audiences, as traditional methods of surveying audience enjoyment, such as questionnaires, provide much less useful data.
The U.S. National Oceanic and Atmospheric Administration (NOAA) is developing a system to help researchers identify and track the endangered North Atlantic right whale using facial recognition algorithms. Only 500 of the whales remain in the world, but tracking them relies on a time-consuming and unreliable method of comparing photos taken during survey flights with a database of photos and drawings of previously documented whales. NOAA partnered with data science challenge company Kaggle to crowdsource the development of a facial recognition algorithm that could reliably identify specific whales based on distinctive markings on their heads. By using this algorithm-driven approach, researchers would have more time to devote to analysis and other projects, rather than spend hours attempting to identify specific whales.
Chinese television news station Dragon TV has begun using an artificial intelligence program called Xiaoice, developed by Microsoft, to deliver weather reports in the place of a human. Though Xiaoice does not analyze weather data itself, it automatically reports official sources and can generate messages related to the weather, such as reminding viewers to wear extra layers when the temperature drops. Microsoft trained Xiaoice, which also operates in China as a smartphone service similar to Apple’s Siri, with real human conversations to make its word choices sound more human and personal.
The Canadian Food Inspection Agency (CFIA) has partnered with the University of Guelph’s Biodiversity Institute of Ontario to develop a method of using genetic markers to help regulators prevent the entry of unwanted and dangerous seafood and plant products into Canada. Genetic information can help CFIA more accurately identify invasive species and mislabeled seafood products that can carry disease or otherwise threaten Canadian food supplies.
The German government, in partnership the Goethe Institut, a cultural organization, and public broadcaster Bayerischer Rundfunk, has created a smartphone app called Ankommen (“arrive” in German) to make it easier for newly arrived refugees to settle in Germany. Ankommen is freely available in Arabic, English, Farsi, French, and German and provides users with German language lessons, job-hunting advice, articles on German culture, and assistance with the asylum process. The government piloted the app in refugee camps and populated it with information based on user feedback, such as information about religious freedoms in Germany.
Image: Merrill Gosho.