CFE 2019: Start Registration
View Submission - CMStatistics
Title: Moment identifiability of homoscedastic Gaussian mixtures Authors:  Carlos Amendola - Technical University of Munich (Germany) [presenting]
Daniele Agostini - Humboldt University Berlin (Germany)
Kristian Ranestad - University of Oslo (Norway)
Abstract: The focus is on the problem of identifying a mixture of Gaussian distributions with same unknown covariance matrix by their sequence of moments up to certain order. The approach rests on studying the moment varieties obtained by taking special secants to the Gaussian moment varieties, defined by their natural polynomial parametrization in terms of the model parameters. When the order of the moments is at most three, we prove an analogue of the Alexander-Hirschowitz theorem classifying all cases of homoscedastic Gaussian mixtures that produce defective moment varieties. As a consequence, we determine identifiability when the number of mixed distributions is smaller than the dimension of the space. In the two component setting we provide a closed form solution for parameter recovery based on moments up to order four, while in the one dimensional case we interpret the rank estimation problem in terms of secant varieties of rational normal curves.