I need help with the following question:

Consider m observations $(y_1; n_1); ... ; (y_m; n_m)$, where $y_i \sim Bin(n_i; θ_i)$ are binomial variables. Assume that $θ_i \sim w_1Beta(α_1; β_1) + w_2Beta(α_2; β_2)$ are mixture from two Beta distributions $(w_1 + w_2 = 1)$
- Derive a Laplace approximation of the likelihood and each mixture component of the prior.
- Derive the empirical Bayes likelihood of the data by integrating out $\theta_i$ using the Laplace approximation, and leave the hyper-parameters $(w_j ; α_j ; β_j)$ (j = 1; 2)
- Derive the EM algorithm to estimate the hyperparameters (you may use also mixture of Gaussian prior to approximate mixture of Beta prior)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.