This week’s list of data news highlights covers July 3, 2021 – July 9, 2021 and includes articles on mapping the shape of the universe and launching smart e-scooters.
Researchers at the University of Western Australia have created an AI system that can classify galaxy shapes within seconds. The team trained the system on images from over 14,000 galaxies. In tests, the system classified elliptical, lenticular, and spiral galaxies with 83 percent accuracy and distinguished between elliptical and spiral galaxies with 97 percent accuracy.
2. Deploying Supercomputers in Pharmaceutical Research
Nvidia has deployed a new AI supercomputer focused on pharmaceutical research. The supercomputer, known as Cambridge-1, is the most powerful supercomputer in the United Kingdom. Researchers from pharmaceutical companies AstraZeneca and GlaxoSmithKline plan to use the supercomputer to study the human genome, model diseases, and discover new drug compounds.
3. Predicting Alzheimer’s Disease
Researchers at Michigan State University have created an AI system that can identify early signs of Alzheimer’s disease by catching disparities in people’s speech and vocabulary. The team trained the system on linguistic patterns and facial expressions from clinical trials. Preliminary tests of the system show that it is as accurate as MRIs in detecting early signs of Alzheimer’s disease.
4. Interacting with the Home Run Derby
Major League Baseball officials have developed an augmented reality app for fans to engage with the upcoming Home Run Derby. Upon opening, the app instructs fans to scan their surroundings for a flat surface. Once a surface is identified, a virtual stadium shows data and animated replays from the 2019 Home Run Derby, and a countdown clock to this year’s Home Run Derby on July 12th.
5. Revealing the Shape of the Universe
Astronomers at the National Astronomical Observatory of Japan have developed an AI system that can more accurately map the shape of the universe by clarifying distortions in distant galaxies. The team trained the system on astronomical data from Japan’s Subaru Telescope and 25,000 mock galaxies created on an astronomy supercomputer in Japan. Using astronomical data, the system revealed previously unobservable details and produced a cosmological model that is consistent with the expected shape of the universe.
6. Increasing Transparency in City Data
City officials in Stockton, California have launched an open data portal for residents to access municipal data. The portal contains 23 datasets with information on public works, public safety, and city development. Residents can use the portal to view municipal data and create customized data visualizations.
Researchers at the University of California, Davis have developed a machine learning model that can identify meteorites after they have fallen to Earth. The team trained the model on images of their collected meteorites in various terrains and meteorites from the Internet. In tests, the model successfully distinguished meteorites from rocks in pictures of a known meteorite fall near Walker Lake, Nevada.
Voi, a Swedish e-scooter operator, and Luna Systems, a micromobility start-up based in Ireland, have partnered to launch a smart e-scooter. The scooter uses sensors to collect data on roadway conditions and nearby pedestrians. Then, they transmit the data to an algorithm that classifies the surrounding environment and issues safety alerts if it detects dangerous conditions. In early tests, the algorithm identified road surfaces and pedestrians with 90 percent accuracy.
9. Identifying Human Influences in Extreme Weather
Earth scientists at the University of California, Los Angeles have developed a machine learning model that can identify human influences in extreme weather events. The team trained the model on data about the largest single-day precipitation events from 1982 to 2015. The researchers then used the model to compare 11 global land precipitation records from the same years and found human influences in every record.
10. Streaming Music Around the Clock
Uncanny Valley, a music technology company in Australia, has developed an AI system that can produce a stream of AI-generated music non-stop. The team trained the system on music data from 200 Eurovision songs and 60,000 lines of lyrics from U.K. Eurovision submissions. The system samples songs from the AI Song Contest, a Eurovision-inspired musical contest, and turns them into a stream of music that plays continuously.
Image credit: NASA Hubble Space Telescope