Chayes' early work focused on the mathematics of phase transitions, both in physical systems like spin glasses, and in combinatorics and computer science. Some of Chayes' later work uses statistical physics approaches to explain the effectiveness of deep learning. Chayes is best known for her work on network science, from mathematical modeling of networks, to algorithms on networks, to machine learning of networks, and finally to applications of network models and algorithms to economic, social, and biological processes. Much of Chayes' work concerns graph limits, a field she cofounded; these are continuum limits of graphs or networks, similar to thermodynamics as a limit of statistical physics, or differential equations as a limit of interacting particle systems. Graph limits are now widely used for non-parametric machine learning of large-scale networks. More recently, Chayes has studied machine learning broadly defined, including applications of machine learning to biomedicine, algorithmic fairness, privacy, and climate change. |