A1642
Title: Regularized estimation for locally Cauchy Levy regression
Authors: Yuma Uehara - Kansai University (Japan) [presenting]
Abstract: A continuous-time regression model driven by a locally Cauchy Levy process with high-frequency samples is considered. For this model, Cauchy quasi-likelihood estimation enables estimation of the mean and scale parameters over a fixed time period. However, the estimating function is not convex, and thus estimation accuracy becomes unstable, especially when the dimension of covariate processes is high. As a solution to this problem, a regularized estimation scheme is proposed, and its theoretical properties are derived.
