Annealed Important Sampling for Models with Latent Variables

摘要: This paper is concerned with Bayesian inference when the likelihood analytically intractable but can be unbiasedly estimated. We propose an annealed importance sampling procedure for estimating expectations respect to posterior. The proposed algorithm useful in cases where finding a good proposal density challenging, and estimates of marginal are required. effect estimation investigated, results provide guidelines on how set up precision order optimally implement procedure. methodological empirically demonstrated several simulated real data examples.