Advertisement

Two La Jolla students are chosen as Regeneron Science Talent Search scholars

The Bishop’s School students James Hou and Shirley Xu are among the Regeneron Science Talent Search scholars for 2023.
(Provided by James Hou and Shirley Xu)

Bishop’s School seniors James Hou and Shirley Xu’s research projects put them in the competition; 40 finalists will be named Jan. 24.

Share

James Hou and Shirley Xu, seniors at The Bishop’s School in La Jolla, are Regeneron Science Talent Search scholars, among 300 selected from 1,949 applicants across the United States and China with an eye on streamlining scientific processes.

The Regeneron Science Talent Search, billed as America’s “oldest and most prestigious science and math competition for high school seniors,” asks applicants to submit an original, independent research project, essays and recommendations.

The 300 scholars will be awarded $2,000 each, and their schools will get $2,000 for each enrolled scholar.

Forty scholars will be named Regeneron STS finalists on Tuesday, Jan. 24. The finalists will compete for more than $1.8 million in awards March 9-15 in Washington, D.C.

Here’s a look at James and Shirley:

James Hou

James Hou's Regeneron project uses artificial intelligence and social media to try to better predict earthquake casualties.
James Hou’s Regeneron Science Talent Search project uses artificial intelligence and social media in an effort to improve earthquake casualty predictions.
(Michael Spengler)

James, 17, made it to the list of 300 for his project “Near Real-Time Seismic Human Fatality Information Retrieval from Social Media with Few-Shot Large-Language Models.”

The project uses “the power of [artificial intelligence] and social media in this modern digital age to improve earthquake casualty predictions,” James said.

He collaborated with the U.S. Geological Survey “to measure or predict how many people are going to suffer injuries or deaths” from earthquakes.

James said the project takes the “powerful information source of social media, where people are constantly talking, and processes it with AI and modern technologies in order to give a more accurate and timely response. … This is really useful for lifesaving aid administered by government agencies.”

It’s often difficult to get human surveyors to earthquake sites to search for people injured or dead, James said, so “social media provides a quick window into these sites where people can be reporting on their own phones at a much faster rate than going through official channels that can often be burdened by a lot of this human labor that is required.”

USGS has found that making use of social media in that manner can be difficult, as “you have people tweeting everything and anything,” James said.

“That’s where AI comes in,” he said. “I have, throughout my past research experiences and through robotics, acquired quite a good understanding of AI and its abilities. I leveraged my knowledge to bridge this gap.”

James was part of a team that took second place last year at the First World Championship in robotics.

“I’ve had a passion for AI since ninth grade,” he said.

The idea for his Regeneron project came from his work over the summer at Stony Brook University’s Simons Summer Research Program in New York.

He said his mentor there, Susu Xu, “was already thinking of using AI in earthquake response. … I had a lot of experience working with this type of model that dealt with language [and] social biases present in language AI, so that helped me really take on this project.”

Being selected as one of the top 300 for the Regeneron competition is “a huge honor,” James said. “It’s a great recognition for this project that I’m really proud of.”

As he waits to see whether he’s one of the 40 finalists, he continues to collaborate with his mentor and is “working to integrate our system into the actual USGS response models.”

James credits Bishop’s computer science instructor Marcus Jaiclin with helping to launch his research and learning, and he hopes to apply AI in civil engineering after he leaves the school.

Shirley Xu

Shirley Xu's project found a novel algorithm for the Closest String Problem.
(Michael Spengler)

Shirley, 17, was selected for her project “A Heuristic Solution to the Closest String Problem Using Wave Function Collapse Techniques.”

The Closest String Problem, or CSP, is part of the “NP-Hard/NP-Complete problem family, regarded by mathematicians and computer scientists as impossible to solve perfectly within practical time limits,” she said.

But CSP, a theoretical computer science computational problem that tries to find the geometrical center of a set of input strings, has “promising applications in fields like drug discovery and gene coding,” Shirley said, so she was determined to seek better solutions.

“I developed a novel algorithm for the Closest String Problem and demonstrated that it outperforms multiple state-of-the-art algorithms in previous publications that attempt to solve CSP,” she said.

Shirley said the idea for her project came from video games, which are among her hobbies.

“My algorithm utilizes techniques inspired by Wave Function Collapse, or WFC, a technique used in video games to create complicated visuals,” she said. “Drawing inspiration from Wave Function Collapse, I was able to create a new solution and way of approaching the Closest String Problem.”

Shirley said she’s honored to be among the 300 Regeneron scholars for 2023 and added that it “is extremely fulfilling and encouraging.”

“[I’ve] always enjoyed engaging with others from various disciplines and backgrounds, especially in research,” she said. “To have this opportunity and to share my work as a Regeneron STS scholar is a dream come true.”

Since publishing her work in a peer-reviewed journal, Shirley has received inquiries from several researchers around the world interested in CSP.

“It’s super exciting for me to be able to discuss and collaborate with other researchers in the field,” she said. “My future goal is to continue being a creator and innovator, enabling technology to make bigger impacts.” ◆