This week’s list of data news highlights covers January 15, 2022 to January 21, 2022 and includes articles on building 99 percent error-free quantum processors and using a machine learning model to solve a 243-year-old puzzle.
Arizona State University has announced a new global education initiative to enroll 100 million students around the world in online courses by 2030. Professors will appear as virtual avatars and use an AI system to supplement courses and grade students.
2. Training Machine Learning Models
Researchers at Meta have created an algorithm that can train machine learning models to recognize images, text, and speech at the same time. Their algorithm improves upon current efforts to train machine learning models by unifying training techniques for all three skills.
Amazon is opening its first brick-and-mortar clothing store in Glendale, California. The store will utilize an AI system to provide personalized recommendations on style and fit to customers and send a selection of clothing based off of the recommendations to customers’ dressing rooms.
4. Predicting Cosmic Characteristics
An undergraduate student at Princeton University has created a machine learning model that can accurately estimate the amount of matter in a computer-simulated universe by analyzing just one of its galaxies. A team of researchers led by astrophysicists at the Flatiron Institute in New York confirmed that the model can predict a universe’s density of matter to within ten percent.
5. Estimating Mortality Risks from COVID-19
Researchers at Charity-University Medicine in Berlin, Germany have created a machine learning model that can predict health outcomes for patients hospitalized with COVID-19. The model measures the levels of 14 proteins in a blood sample to provide an estimate of mortality risks.
6. Building Quantum Processors
Three teams of researchers in Australia, Japan, and the Netherlands have built silicon-based quantum processors that are more than 99 percent error-free. Using silicon makes the quantum devices more compatible with other technologies necessary for building quantum computers, and improves their scalability.
An international team of researchers has solved Euler’s 243-year-old combinatorics puzzle with a machine learning model. The puzzle, which is called the 36 officers problem, asks if it is possible to arrange 36 officers drawn from 6 different regiments, each of which also have officers of 6 different ranks, in a square so that each row and column has officers of different rank and regiment. The team used a machine learning model to prove that a solution could theoretically exist in a quantum state.
8. Identifying Anne Frank’s Betrayer
An international team of researchers has created a new hypothesis of who betrayed Anne Frank and her family. The team used an AI system to identify connections between evidence held in archives around the world and found an overlooked note with the potential informant’s name.
9. Buying Jerseys with Augmented Reality
Snapchat has partnered with Microsoft, the Los Angeles Lakers basketball team, and Hexa, a 3D visualization company in Israel, to sell Lakers jerseys on their platform. Users can try on and purchase jerseys virtually with augmented reality.
10. Improving Diagnostic Tools
Researchers at Duke University have created an AI system that can detect potentially cancerous lesions in images of mammography scans and predict whether the patient should undergo further testing. Unlike similar tools, the team trained the system to be interpretable so doctors can understand why the system recommends certain actions.
Image credit: Flickr user Dano