Florentina Bunea, professor of statistics in the Cornell Ann S. Bowers College of Computing and Information Science, has received the Institute of Mathematical Statistics’ Medallion Award, one of the most prestigious awards in the field of statistics.
As part of receiving the IMS Medallion Award, Bunea will deliver a lecture at the Joint Statistical Meetings (JSM) in August in Nashville, Tennessee.
One of eight recipients, Bunea was recognized for fundamental contributions to uncertainty quantification for identifiable latent structures in complex systems. In simple terms, her research helps bring clarity to the hidden workings of complex systems by analyzing the data they generate. A pioneer in statistical machine learning theory and high-dimensional statistical inference, Bunea has had a broad impact in diverse fields, including most recently, to biological discoveries from modern multiomic data.
“This type of research – which spans the entire spectrum, from challenging applications to new theory development – is always collaborative in nature,” Bunea said of receiving the honor. “I share this medallion with all of the outstanding collaborators and Ph.D. students I had the pleasure of working with over the years.”
In the early aughts, motivated by advances in genome sequencing technologies, Bunea began her research career in the then-emerging area of high-dimensional statistics. She and her collaborators coined the term “sparsity adaptive estimation” – a new statistical principle introduced at the time, to guide researchers toward the meaningful signals underlying the generation of massive data sets. Today, its use in either practice or theory is ubiquitous.
In the last decade, Bunea’s research has shifted toward the foundations of interpretable statistical machine learning. The new challenge, she said, is to use theoretically grounded strategies to evaluate data generated by AI models. Adding to their complexity, these models can be “black boxes,” shielded behind proprietary laws and thus potentially off limits for deeper analysis by outside researchers.
“However, if you have access to the system’s outputs, you can get a sense of how it works,” Bunea said. “This is a quintessential statistical problem: find and understand the underlying latent mechanism from noisy data.”
She is currently developing new methods and theory for better understanding outputs from large language models.
Bunea was selected as a fellow of the IMS, is a recipient of the Bowers Research Excellence Award, and has served as an associate editor for many of the field’s most prestigious journals, including the Annals of Statistics, Bernoulli, and the Journal of the American Statistical Association (JASA). Bunea received a Ph.D. in statistics from the University of Washington in 2000 and joined Cornell’s Department of Statistics and Data Science in 2011.
By Louis DiPietro, a writer for the Cornell Ann S. Bowers College of Computing and Information Science.