This week’s list of data news highlights covers June 13-19, 2020, and includes articles about teaching an AI system to stitch wounds and creating an open data platform for autonomous vehicles.
1. Measuring Discrimination on Airbnb
Airbnb has created a project to measure racial discrimination using data from its platform. The study is using data about or from reservations, reviews, and customer support interactions to measure the level of discrimination hosts and travelers face. The project is using profile photos and names to discern what race an individual might associate with a host or traveler.
2. Teaching an AI System to Stitch Wounds
Researchers from the University of California, Berkeley, and Intel have developed an algorithm that can mimic the motions required for stitching wounds and surgical incisions. The researchers trained the algorithm using videos that contained 80 examples of surgeons using a surgical robot, programming the algorithm to observe actions such as needle insertion and threading movements. The algorithm successfully mimicked the motions when it controlled a surgical robot.
3. Creating an Open Data Platform for Autonomous Vehicles
The U.S. National Highway Traffic Safety Administration has launched an open data initiative for autonomous vehicle developers and states to share data. Nine firms are participating in the effort, including firms such as Waymo and Uber. Eight states are participating, and participating groups will share data from the on-road testing and development of autonomous vehicles systems.
Researchers from the Oregon Health & Science University have developed an AI-enabled system that helps people with type-1 diabetes manage their glucose levels. The system uses data from glucose monitors and wireless pens to recommend daily insulin dosage levels. Physicians considered 99 percent of the recommendations it provided over 100 weeks of patient testing safe.
5. Using AI for Social Distancing
Cognize, a technology firm based in California, has developed a platform that uses IoT-data and AI to help businesses follow social distancing guidelines. The system uses data from video and thermal sensors to detect if individuals are within six feet, wearing a mask, or have a fever. The platform can reduce the need for staff to perform temperature checks or identify individuals not wearing a mask.
Airports in the United Kingdom are using AI to detect drones. Radar systems can struggle to differentiate drones from other small objects such as birds, so Bristol Airport in England is using a system that can analyze the signal outputs and movement profile of an object to detect if it is a drone. London’s Heathrow Airport uses an AI-enabled system to detect drones in images.
7. Bringing the NBA Back Safely
The National Basketball Association is providing players smart rings that can help detect unwell individuals. Oura, a wellness wearable company based in Finland, designed the rings to measure body temperature, heart rate, and respiration rate. Researchers from the University of Michigan will analyze the data to generate a wellness assessment report for each player.
8. Using AI to Make Investments
Swedish venture capital firm EQT Ventures has used an AI system called Motherbrain to help the organization invest more than $100 million. Motherbrain uses convolutional neural networks to analyze time-series performance data about more than 10 million companies, helping EQT Ventures prioritize the firms it should provide funds. The system allows EQT Ventures to find potential investments outside of its network of connections.
A German-Russian space telescope has created a map of where matter is accelerating, heating, and shredding in the universe. Matter in space, such as black holes, stars, and supernova remnants give off x-rays, and the telescope has detected over a million sources of x-rays. The orbiting telescope is 1.5 million kilometers from Earth, and the map it created includes massive black holes that formed when the universe was less than ten percent of its current age.
10. Deep Learning Without Specialized Hardware
Neural Magic, a startup based in Massachusetts, has developed software that allows deep learning algorithms to perform faster on central processing units (CPUs). Many AI-enabled systems use graphics processing units (GPUs) in conjunction with CPUs because the former can process calculations faster. However, GPUs often store much of the data they process on another chip because they have limited cache memories. Neural Magic has redesigned deep learning algorithms to use a CPU’s large available memory and complex cores to save time by eliminating the need to move data on and off the chip.
Image: Chensiyuan