![[Artificial Intelligence and Statistics Logo]](/aistats2022/assets/images/logo-aistats.jpg){kind=logo}
[edit]
AISTATS 2022 Invited Talks
Day 1: Mon, March 28
New Methods for Taming Endogeneity Bias in Observational Studies: Proxies, Bespoke Instruments and Invalid Instruments
Eric J. Tchetgen Tchetgen
Abstract
This talk will discuss recent innovations in causal inference literature on the identification and estimation of causal effects from observational data in presence of endogeneity or equivalently unmeasured confounding bias. We will focus primarily on three recent developments: (i) The Proximal causal inference framework that leverages imperfect proxies of unmeasured confounders to remove hidden bias in observational analyses; (ii) Bespoke Instrumental variable framework that leverages a reference population in which a known intervention sets the treatment to generate a bespoke instrument tailored to account for endogeneity in a target population of interest; (iii) Invalid instrumental variable framework that leverages one or more invalid instruments to nevertheless correct for endogeneity bias without requiring that core instrumental variable assumptions hold. We view these new techniques as important strategies to relax the standard un-confoundedness assumptions commonly used in practice. Machine Learning tools implementing the methods and corresponding small bias guarantees will be described along with several empirical examples demonstrating the new methods in action.
Day 2: Tue, March 29
Machine Learning for Ocean and Climate Modeling: Advances & Challenges
Laure Zanna
Abstract
Climate simulations remain our best tools to predict global and regional climate change. Climate projection uncertainty stem, in part, from the poor or lacking representation of processes, such as turbulence, clouds that are not resolved on the grid of global climate models. The representation of these unresolved processes has been a bottleneck in improving climate projections. The explosion of climate data and the power of machine learning algorithms are suddenly offering new opportunities. For example, can data-driven machine learning methods help us deepen our understanding of these unresolved processes and simultaneously improve their representation in climate models to reduce climate projections uncertainty? In this talk, I will discuss the current state of climate modeling and its future, focusing on the advantages and challenges of using machine learning for climate projections. I will present some of our recent work in which we leverage tools from machine learning and deep learning to learn representations of unresolved ocean processes and improve climate simulations. Our work suggests that machine learning could open the door to discovering new physics from data and enhance climate predictions. Yet, many questions remain unanswered, making the next decade exciting and challenging for hybrid climate modeling.