Details
Speaker: Rose Cersonsky, Ph.D.
Michael and Virginia Conway Assistant Professor
Chemical and Biological Engineering
University of Wisconsin-Madison
Title: “Extending the successes of atomistic machine learning up a scale or two”
Abstract: Machine learning has delivered remarkable results in atomistic simulations, enabling predictive accuracy at scales previously unreachable. However, extending these successes to the nanoscale requires addressing several fundamental obstacles. First, the generation of atomistically resolved training data at larger scales is prohibitively costly. Second, retaining full atomic resolution in models becomes impractical at extended time and length scales, demanding strategies for coarse-graining without losing essential physics. Finally, the characterization of phases and behaviors in systems with glassy or gel-like morphologies poses unique challenges: these materials lack sharp structural boundaries, evolve slowly, and exhibit complex dynamical heterogeneity, making them difficult to quantify and model. In this talk, I will explore these challenges, outline emerging strategies to overcome them, and discuss how machine learning frameworks can be adapted to bridge the gap between the atomistic and mesoscale worlds.
Bio: Rose K. Cersonsky is the Michael and Virginia Conway Assistant Professor of Chemical and Biological Engineering at the University of Wisconsin-Madison. She received her Bachelor of Science degree in materials science and engineering with a minor concentration in computer science from the University of Connecticut in 2014. She went on to obtain her Ph.D. in macromolecular science and engineering from the University of Michigan in 2019 alongside Professor Sharon C. Glotzer, focusing on the self-assembly behavior and optical properties of colloidal nanoparticles. Following her doctoral work, she collaborated with Prof. Michele Ceriotti as a postdoctoral researcher at École Polytechnique Fédérale de Lausanne in Lausanne, Switzerland, working on developing and applying hybrid supervised-unsupervised machine learning models for data-driven studies of molecular design.
Rose’s research group at UW-Madison, established in 2023, centers on developing techniques for and using data science and machine learning to unify our understanding of molecular motion and emergent phenomena across length scales. The Cersonsky Lab specializes in information-theoretic approaches to machine learning in multiscale systems, particularly how representation choice controls model expressivity. Our work has shown that careful attention to data representations can allow simple, interpretable models to outperform complex architectures, especially in data-limited or noisy datasets.
She and her group lead the development of scikit-matter, a scikit-learn-affiliated package for quantitative structure-property relations in materials research, and are core developers of chemiscope, an interactive visualizer for data-driven analyses of molecular datasets. Rose’s work has been recognized with a number of awards, including being named one of Matter’s “35 under 35” in Materials Research, an Emerging Investigator by the International Association for Colloids and Interface Scientists, the Victor K. LaMer Award from the Colloids Division of the American Chemical Society, the Biointerfaces Institute Innovator Award, and the Charles G. Overberger Award for Excellence in Research.
In addition to research, Rose has devoted herself to scientific service, leading and coordinating multiple outreach programs and publishing work in educational journals on community engagement and gender equity. Recently, she released the commentary “Not yet defect-free: the current landscape for women in computational materials research,” in npj Computational Materials, highlighting the data-driven inequities still demonstrated by the field.