This week’s list of data news highlights covers December 13-December 23, 2019, and includes articles about Russia launching a quantum computing initiative and California allowing firms to test autonomous trucks on its public roads.
1. Identifying Heart Failure During a Standard Physical Exam
The Mayo Clinic and Eko, a firm that creates digital health products, have developed a system that uses AI and a digital stethoscope to detect heart failure in a standard physical exam. The system analyzes 15 seconds of electrocardiogram data from a digital stethoscope to make its analysis. When tested on the data of nearly 53,000 patients, the system had an accuracy of almost 86 percent.
Researchers from Uber AI have developed a new method that helps AI models that generate text stay on topic. Uber’s technique uses two models, one which generates text, and a second model that scores the first model’s output based on the number of relevant words it contains concerning a desired topic or sentiment, such as healthcare and positivity. The first model checks its score after each word to generate appropriate sentences.
3. Russia is Launching a Quantum Computing Initiative
Russia has launched a national quantum computing initiative. The government will provide nearly $800 million in funding to its leading laboratories over the next five years for basic and applied quantum research. The funding is a part of a more than $4 billion research and development investment in digital technologies.
Researchers from Facebook have developed a neural network that can calculate integrals and solve differential equations—other neural networks can only perform addition or multiplication. The researchers trained the network on millions of equations, teaching the network to recognize patterns in the equations. The network often outperformed conventional mathematical packages, such as Matlab, at finding solutions in both accuracy and speed.
5. Increasing Government Accountability With Data
Seattle has launched a public data dashboard that tracks its performance delivering critical city services to increase accountability and integrate siloed city data. The dashboard includes data on seven priority areas, including basic city services, capital projects, and safe and healthy communities. The dashboard displays the cities performance in meeting targets across these areas, such as repairing 80 percent of potholes within 30 days and having a first responder on the scene within four minutes, 90 percent of the time.
6. Advancing the Commercial Use of Autonomous Vehicles
The California Department of Motor Vehicles has announced a permitting process that will allow firms to test or deploy autonomous trucks for commercial use on its public roads. The rule applies to trucks that are 10,000 pounds or less, which includes utility vans. The new rulemaking can help advance the commercial use of autonomous vehicles for food or package delivery.
7. Using an AI System to Recommend Cars
Kar Auction Services, a vehicle auctioneer, is using an AI-enabled system to helps its customers—car dealerships—identify which vehicles are likely to sell fastest and generate the highest profit in a market. The system uses data such as a vehicle’s condition, make and model, and the selling history of a dealership to make its recommendations. Since its implementation in March of 2019, the system has led to tens of thousands of transactions.
8. Identifying Soldiers in Civil War Photos
Researchers led by an individual from the Virginia Polytechnic Institute and State University have developed a web-based tool called Civil War Photo Sleuth that uses facial recognition to identify soldiers in photos. The tool’s facial recognition technology uses 27 facial landmarks and the distance between them to analyze the similarity between photos of unknown and known individuals. Since its inception in 2018, the tool has helped make more than 3,000 identifications.
Google has developed Wildlife Insights, a tool that uses AI to help conservationists track animals. The tool allows users to upload their camera trap images, and the tool will automatically discard images without animals and label the photos with animals by species. The tool also provides the public access to 4.5 million images, which individuals can filter by species and location to track wildlife populations.
MetaFact, a startup in India, has developed a tool that uses natural language processing to identify potential misinformation on news websites, blogs, and social media platforms. The tool attempts to identify the context of sentences and flags sentences with declarative or interrogative tones that may contain misinformation, and then journalists review the claims.
Image: Shadow Dragon