B1371
Title: Online inference for tensor models
Authors: Wei Sun - Purdue University (United States) [presenting]
Abstract: An online tensor model is considered where the true model parameter is a low-rank tensor and proposes a fully online procedure to make sequential decision-making and conduct statistical inference simultaneously. The low-rank structure of the model parameter and the adaptivity nature of the data collection process make this difficult: standard low-rank estimators are not fully online. They are biased while existing inference approaches in online models fail to account for the low-rankness and are also biased. To address these, a new online doubly-debiasing inference procedure is introduced to handle both sources of bias simultaneously. In theory, the asymptotic normality of the proposed online doubly-debiased estimator is established, and the validity of the constructed confidence interval is proven.