PASS-GLM: polynomial approximate sufficient statistics for scalable Bayesian GLM inference

  Abstract