A1167
Title: Optimal backward-learning approach for Gaussian linear structural equation models
Authors: Gunwoong Park - Seoul National University (Korea, South) [presenting]
Abstract: The first optimal backwards-learning approach for Gaussian linear structural equation models (SEMs) is introduced using the best-subset-selection approach. Specifically, the class of optimally identifiable Gaussian linear SEMs is provided. Subsequently, it proves that the proposed algorithm is optimal in terms of the sample complexity. Various simulations verify the theoretical findings and confirm the outstanding performance of the algorithm.
