Ella Lu ’26, one of the 300 Scholars in the Regeneron Science Talent Search, knew that the top 40 Finalists for the nation’s oldest and most prestigious science research competition for high school students would soon be announced. She wasn’t expecting to receive a direct call from them, however. So when the Society for Science called her as she waited in the Raleigh-Durham airport to board the first leg of a January Term flight to Buenos Aires, she thought the worst. 

“They were like, ‘Oh, guess what we’re calling about,’” Ella recalls. “And I honestly didn’t know, so I was like, ‘I have no idea.’ I thought maybe there was an error in my project and they were going to revoke me getting top 300.”

Not hardly. The representative for Society for Science, which operates the Regeneron Science Talent Search, was calling to tell Ella she had made the Top 40. Not bad at all, considering 2,600 students entered their research into the academic competition.

The research for which Ella is being recognized focuses on utilizing AI to analyze visual art. She created a program called Compositional Analysis of Visual Art Structure (CANVAS), which uses machine learning and modern computational techniques to analyze composition in traditional Impressionist paintings. The point, Ella says, is to train AI to evaluate the technical merits of art in a way only humans can, searching for patterns and commonalities among highly-valued works, but also in a way that can be scaled up to large-scale analyses of artistic work.

The idea for the research sprang from a moment during her sophomore year at Chapel Hill High, when, as a devoted painter, she was introduced by her AP Art teacher to a number of compositional techniques she and other students in the class could employ in their paintings as they developed their skill.

“I began to wonder if this is actually something that artists across centuries actively all thought about. Was this [the types of compositions] something that they have in common? I wanted to see if there are patterns throughout history. I guess the tool that I built is a way for us to continue studying that.”

Though this was an independent project that Ella worked most intensely on in the summer between her junior and senior years at NCSSM-Durham, she credits NCSSM and its research programs – particularly the Research in Computational Science course – for providing her with the opportunity to explore ideas that inspire her the most.

“Science and Math has allowed me to do research that I’m truly passionate about and ask the questions that I’m curious about. You can’t really find that anywhere else. But at NCSSM, you can truly be original and do what you want to do.”

Ella will travel to Washington, D.C. in early March for Finals Week, where she will get to meet other finalists and share her work publicly. Winners of the competition will be announced March 10.

The Regeneron Science Talent Search is a program of Society for Science, a D.C.-based 501(c)(3) membership organization that has been expanding scientific literacy, access to STEM education and scientific research for more than 100 years. The organization is led by NCSSM alum Maya Ajmera ’85, President and CEO. She is also the Executive Publisher of Science News, the organization’s award-winning magazine.

---