报告题目： Identification of Causal Relationships with Latent Variables: an Overview
报 告 人: Biwei Huang
In many instances, we are unable to measure all task-related variables, leading to the emergence of a hidden causal world. Exploring causal relationships in the presence of latent variables, as well as those among latent variables, is not only essential for understanding and interpretability but also facilitates various downstream tasks. In this talk, I will provide an overview of recent advancements in identifying causal relationships with latent variables. Specifically, I will categorize recent advancements into two groups: one leveraging the strategy of distribution shifts to achieve identifiability, and the other relying on structural sparsity. In the former category, representative works primarily focus on causal representation learning, while in the latter category, representative works are predominantly related to latent causal structure discovery.
Biwei Huang is an assistant professor at the University of California San Diego. She received her PhD degree from Carnegie Mellon University, under the supervision of Prof. Kun Zhang and Prof. Clark Glymour. Her research interests are mainly in three aspects: (1) automated causal discovery in complex environments with theoretical guarantees, (2) advancing machine learning from the causal perspective, and (3) using or adapting causal discovery approaches to solve scientific problems. On the causality side, Huang's research has delivered more reliable and practical causal discovery algorithms by formulating and addressing the property of distribution shifts and allowing nonlinear relationships, general data distributions, latent confounders, etc. On the machine learning side, her work has shown that the causal view provides a clear picture for understanding advanced learning problems and allows going beyond the data in a principled, interpretable manner.