This course will explore recent work that leverages machine learning (ML) as a tool for tackling climate change, with a focus on climate science and climate adaptation. We will discuss how modern machine learning can be used to assess, understand and respond to projected climate extremes, natural disasters, and environmental change. The target audience for this course is advanced undergraduate students or graduate students who are interested in using ML and AI to address high-impact global issues. Students will read and discuss recent research papers on ML for Climate and complete an original project as a member of a multidisciplinary team.
Climate themes may include: Climate models and predictions; Extreme weather and natural disasters; Farms and forests; Oceans and marine ecosystems; Climate misinformation.
Machine learning topics may include: Physics-informed learning and emulators; Explainable AI; Uncertainty quantification; Image super-resolution; Graph neural networks, Policy optimization.
This course is offered jointly by the Department of Earth, Environmental and Planetary Sciences and the Brown Data Science Initiative.
EEPS-DATA 1720 is a seminar course centered around presentations and discussions of recent research papers. Introductory lectures in the first weeks of the course will introduce students to basics of climate change, with additional mini-lectures given by the instructor throughout the course. Students will work in interdisciplinary teams on a course project.
One of the following is required: Familiarity with fundamentals of machine learning through coursework (e.g. CSCI 1420, CSCI 1470, DATA 1030, EEPS 1960D), research or project experience, OR a background in climate science (e.g. EEPS 0850, 1430, 1510 or 1520) and permission of the instructor.
Basic programming experience in Matlab, Python, R, or any high-level programming language is required.
Students should have an interest in climate science, adaptation and mitigation, but prior coursework in Earth and/or environmental science is not necessary.
The instructor of this course also offers a Spring semester course (not offered this year) called EEPS 1960D: Machine Learning for the Earth and Environment. EEPS 1960D is an introductory data science course geared toward students in the Earth and physical sciences. EEPS 1960D assumes no prior knowledge of machine learning.
EEPS-DATA 1720 is course designed for students with prior experience in data science (see prerequisites or consult the instructor) who are looking to deepen their knowledge of machine learning tools and learn about how these tools can advance climate research.
Note: in the future EEPS 1960D will be offered under a different course number (tentatively EEPS 1340)