Causal Discovery with Hidden Variables Based on Non-Gaussianity and Nonlinearity

A central problem of science is to elucidate the causal mechanisms underlying natural phenomena and human behavior. Statistical causal inference offers various tools to study such mechanisms. However, owing to the lack of background knowledge, it is often difficult to prepare causal graphs required for performing statistical causal inference. To alleviate the difficulty, we have worked on developing statistical methods for estimating causal relationships from observational data obtained from sources other than randomized experiments and constructing a new methodology that goes beyond the conventional limits. This chapter provides an overview of recent developments in our work and other relevant work. In particular, we focus on hidden variable models, nonlinear models, and mixed data models.