Robust Bayesian Inference in l q -Spherical Models
本文引入了一类多元l_q-球面分布，证明在尺度参数的无信息先验下，位置参数的后验推断对所有具有相同q的l_q-球面采样模型都相同，从而实现了对指数幂分布独立采样偏离的完美稳健性。
The class of multivariate I4-spherical di,tributions is introduced and defined through their isodensity surfaces.We prove that, under a JeHreys' type improper prior on the scale parameter, posterior inference on the location parameters is the same for all Iq-spherical sampling models with common q.7itis gives us perfect inference robustness with respect to any departures from the reference case of independcnt sampling from the exponential power distribution.