By Vinod Kumar Chauhan, University of Oxford

Abstract 

Limited dataset sizes can pose challenges in causal inference and machine learning. This is especially problematic in causal inference, where data is split among different treatment groups for model training, potentially leading to bias. While some information sharing among treatment groups can help, current individualized treatment effect (ITE) learners often lack a mechanism for comprehensive inter-treatment information sharing. To tackle this, we introduce a novel deep learning framework for training ITE learners. It leverages dynamic end-to-end information sharing among treatment groups through soft weight sharing of hypernetworks. This framework, referred to as HyperITE, complements existing ITE learners and effectively reduces ITE estimation errors, particularly benefiting smaller datasets in our experiments.