All Times EDT
In this talk, I will present a recent line of research on semiparametric estimation in high-dimensional index models using a generalized Stein's identity. We consider estimation in three different index models: varying index coefficient models, index volatility models, and graph index models. Based on the generalized Stein’s identity, we develop a computationally efficient estimator for the high-dimensional parameters without estimating the link functions. Our estimators are shown to achieve optimal statistical rates of convergence, while requiring covariates to only satisfy weak moment conditions, in contrast to existing literature on sliced inverse regression that assumes the covariates to be Gaussian or elliptically symmetric. Extensive numerical experiments corroborate the theoretical results.