On the Difficulty of Unbiased Alpha Divergence Minimization
Tomas Geffner 1 Justin Domke 1
Abstract Existing alpha-divergence minimization algorithms can be
Several approximate inference algorithms have classified into two broad groups: biased methods (Li &
been proposed to minimize an alpha-divergence Turner, 2016; Regli & Silva, 2018) and unbiased methods
between an approximating distribution and a tar- (Kuleshov & E ...


雷达卡




京公网安备 11010802022788号







