EDS Seminar Series
Machine Learning

Geospatial Machine Learning: Data-focused algorithm design, development, and evaluation

EDS seminar speaker series. Esther Rolf discusses Geospatial Machine Learning: Data-focused algorithm design, development, and evaluation

Title: Geospatial Machine Learning: Data-focused algorithm design, development, and evaluation
Speaker: Esther Rolf

Abstract: Geospatial machine learning (geo-ML) has incredible potential to help organize global data from heterogeneous data sources -- from satellite imagery to citizen science data to ground surveys -- in order to help people make more informed decisions to address challenges like climate change and global health. At the same time that geo-ML models are increasingly sought after for this promise, there are key technical distinctions of geo-ML when it comes to designing, developing, and evaluating useful prediction models. The sheer scale of global data (e.g. satellite images as input to a predictive model) necessitates computationally efficient algorithms in order for predictive models to be accessible to all. Spatio-temporal autocorrelations can make it difficult to evaluate real-world model performance, and can lead to concerns of unfairness for decisions derived from standard geospatial models. The heterogeneity of available geospatial data sources presents a challenging opportunity to design models that account for the structures of data from different sensors and modalities. In this talk, I will discuss why focusing on these interconnected facets of geospatial data -- scale, spatio-temporal autocorrelations, and modal heterogeneity --  is key to achieving effective algorithms and evaluation frameworks for geospatial machine learning

Speaker Bio: Esther Rolf is a postdoctoral fellow with the Harvard Data Science Initiative and the Center for Research on Computation and Society. In fall of 2024, Esther will join CU Boulder as an assistant professor of computer science.

Esther's research in statistical and geospatial machine learning blends methodological and applied techniques to study and design machine learning algorithms and systems with an emphasis on usability, data-efficiency, and fairness. Some of her specific projects span developing algorithms and infrastructure for reliable environmental monitoring using machine learning, responsible and fair algorithm design and use, and the influence of data acquisition and representation on the efficacy and applicability of machine learning systems.

Esther completed her PhD in Computer Science in 2022 at UC Berkeley, where she was advised by Ben Recht and Michael I. Jordan. Esther’s PhD was supported by an NSF Graduate Research Fellowship, a Google Research Fellowship, and a UC Berkeley Stonebreaker Fellowship. Esther has won best paper awards at ICML (2018) and the Workshop on AI for Social Good at Neurips (2019), and the impact of her work has been recognized with a SDG Digital Gamechangers award (2023) from the United Nations Development Programme and the International Telecommunication Union.