Rose Yu: Uncertainty Quantification in Learning Spatiotemporal Dynamics
Abstract: Applications such as public health, transportation, and climate science often require learning complex dynamics from large-scale spatiotemporal data. While deep learning has shown tremendous success in these domains, prior works have mostly focused on point estimates without quantifying the uncertainty of the predictions. In high-stakes domains, being able to generate probabilistic predictions with confidence intervals is critical to risk assessment and decision making. In this talk, I will present our efforts in uncertainty quantification (UQ) in learning spatiotemporal dynamics. I will discuss (1) a systematic study of UQ for deep spatiotemporal forecasting. We analyze UQ methods from both the Bayesian and the frequentist point of view, casting in a unified framework. We perform benchmark tests on COVID-19, traffic forecasting and air quality prediction tasks. (2) Interactive Neural Process (INP), a Bayesian active learning framework that can significantly accelerate stochastic simulation. We demonstrate our method on the use cases of COVID-19 forecasting and scenario creation.
Bio: Dr. Rose Yu is an assistant professor at the University of California San Diego, Department of Computer Science and Engineering. She was a Postdoctoral Fellow at the California Institute of Technology. Her research focuses on advancing machine learning techniques for large-scale spatiotemporal data analysis, with applications to sustainability, health, and physical sciences. A particular emphasis of her research is on physics-guided AI which aims to integrate first-principles with data-driven models. Among her awards, she has won Faculty Research Award from JP Morgan, Facebook, Google, Amazon, and Adobe, Several Best Paper Awards, Best Dissertation Award in USC, and was nominated as one of the ’MIT Rising Stars in EECS’.