Yu receives AFOSR grant

Christina Lee Yu has been awarded a grant from the Air Force Office of Scientific Research for her project “Efficiently Exploiting Structure for Causal Inference in the Presence of Network Interference.”

The three-year, $450,000 award will fund Yu’s research into how best to predict outcomes of an intervention from limited data when there may be complex dependences in the outcomes arising from network interference.

Yu, an assistant professor in Cornell's School of Operations Research and Information Engineering, explains that existing standard data analyses are able to understand and describe data, but are unable to answer “what if” and “why” questions about predicted outcomes of possible interventions. These existing data analyses can leave policy-makers guessing at the effectiveness of proposed changes.

These “what if” and “why” questions are often hard to answer in existing models because of what is known as network interference, where an intervention directed at one subject may actually affect another subject’s outcomes as well. When this is the case, it is difficult to accurately assess the effects of an intervention.

Air Force Office of Scientific Research logoTraditional methods for predicting the effects of policy interventions have a hard time accounting for network interference. This grant from the AFOSR will help Yu work on creating new models that can better take advantage of data structure and lead to more accurate measures of causal inference under conditions of network interference.

Researchers and policy makers in many domains can benefit from a more accurate and reliable approach to the study of causal inference under network interference. When trying to predict the effects of an intervention, public health officials, economists, business strategists, and lawmakers all rely on models to predict the possible outcomes of specific policies or actions. Having better tools to predict these outcomes will lead to better decisions.

“I believe this work is particularly timely.” Yu said. “I am excited about the impact it will have as data-driven decision making is increasingly being incorporated into everyday domain of our society, and causal inference in the presence of complex network dependencies is a key player for guiding data-driven decisions.”

Other Articles of Interest