- This event has passed.
February 2 @ 12:40 pm - 1:40 pm
Title: Causal Inference Methods To Detect And Reduce Societal Inequalities In Data-Driven Applications
Presenter: Razieh Nabi Johns Hopkins University
Abstract: Causal inference helps us identify and address different sources of bias in data analyses, including confounding bias and bias due to systematic censoring and missing values. In this talk, I will focus on a source of bias that is reflected in data through historical patterns of inequality and discrimination against minorities and underserved communities. Such injustices are easily propagated in the system via naive implementations of automated decision making in socially impactful settings like criminal justice and healthcare. In my talk, I describe the framework I have developed based on causal and counterfactual reasoning to detect and counteract such biases in statistical and machine learning models.