This week’s list of data news highlights covers July 25-July 31, 2020, and includes articles about forecasting the movements of locust swarms and mapping 80 percent of the universe.
Researchers from institutes in France, Germany, Portugal, and South Africa have developed an AI system that can identify individual birds. The researchers collected images of the birds using bird feeders equipped with cameras and sensors, and used this data to train the system. The system can identify individual birds of the same species with 87 percent or higher accuracy.
The United Nations Food and Agricultural Organization, an agency within the UN, is using data to forecast the movement of desert locusts. The organization uses a tool that combines real-time reports of locust movements, satellite imagery of vegetation, and weather data to predict where swarms will move next. The swarms destroy crops, threatening the food security of humans populations.
Researchers from more than 50 organizations have created a 3D map that depicts 11 years of history in space. The researchers combined data from multiple telescopic surveys of space to create the map, allowing them to chart 80 percent of the universe’s history. This data helped the researchers measure the rate at which the universe is expanding and confirmed one theory of how the universe evolved from being homogenous to clumpy—some areas contain large galaxies and others containing nothing.
Researchers from Intel, MIT, and the Georgia Institute of Technology have developed an AI-enabled system that can understand code. The system compares a piece of code to millions of other snippets of code it has seen and uses a neural network to find code that performs a similar task. Intel plans to use the system to help its developers find ways to write more efficient programs.
Researchers from the Hefei Institutes of Physical Science, a research center in China, have developed a mobile device that can trace air pollution sources. The device, which attaches to a vehicle, uses a sensor and positioning system to measure volatile organic compounds (VOCs), which are key sources of air pollution. The researchers have deployed the device in several cities across China and are using the data it collects to map VOCs’ distribution in real-time.
Researchers from MIT have developed a machine learning system that can help design and evaluate vaccines that use peptides—short strings of amino acids that stem from a virus and can build an immune response. The system, OptiVax, scores a virus’ different peptides using multiple factors, such as how they react to human genomes from diverse populations. OptiVax then uses this data to design a vaccine that maximizes the number of peptides displayed per individual in different regions.
Researchers from the Max Planck Institute of Animal Behavior and the University of Konstanz, both in Germany, have attached sensors to farm animals to find that the animals increased their activity before earthquakes. The researchers attached the sensors, which captured the animals’ movements, to cows, sheep, and dogs. More than 18,000 tremors occurred during the study period, and the animals’ activity significantly increased before magnitude 3.8 earthquakes or greater when in a confined space.
BMW is using AI to detect defects in vehicles during production. The system analyzes photos of vehicles on an assembly line to detect abnormalities such as missing screws. The software replaces a system that compared each photo of a vehicle to a target image, which led to more false positives as issues such as unwanted reflections confused the system.
Researchers led by an individual from Western University in Canada have analyzed data about over 11,000 relationships to find that characteristics about the relationship better predicted relationship quality than individual traits of the people involved. The researchers used data from 43 datasets and used machine learning to determine the predictive power of variables. They found that the most predictive variables were perceived partner commitment and appreciation for a partner.
The U.S. National Institutes of Health has launched a $58 million initiative to advance data science in Africa. The initiative will fund an open data science platform and coordination center, up to five research hubs, up to four training programs, and ethical, legal, and social implications research. The agency believes the initiative can lead to the development of applications that have impacts both in and outside the continent, such as the development of large datasets that reveal disease patterns.
Image: Adrian Pingstone