DOI: 10.21276/sjpms.2019.6.7.2

Pages: 124-129

The intrinsic estimation for parameters are extensively studied in the literatures, which does not depend on the coordinate systems or model parametrization. In this paper, the intrinsic variance estimation for Gaussian mixture distribution is provided. The optimal estimators are proposed by minimizing the mean squared Rao distance and the mixture of symmetrized Kullback–Leibler divergence between normal distributions. Keywords: Intrinsic estimation, Gaussian mixture distribution, Variance estimation, Rao distance, Kullback–Leibler divergence.