def marginal(guide, num_samples=100):
posterior_predictive = Predictive(guide, num_samples=num_samples)
posterior_samples = posterior_predictive.forward(data)
mu_mean = posterior_samples['mu'].detach().mean(dim=0)
beta_mean = posterior_samples['beta'].detach().mean(dim=0)
weights_mean = mix_weights(beta_mean)
centers, _ = truncate(alpha, mu_mean, weights_mean)
return centers
def marginal(guide, num_samples=25):
posterior_predictive = Predictive(guide, num_samples=num_samples)
posterior_samples = posterior_predictive.forward(data)
mu_mean = posterior_samples['mu'].detach().mean(dim=0).squeeze()
prec_mean = posterior_samples['prec'].detach().mean(dim=0).squeeze()
corr_chol_mean = posterior_samples['corr_chol'].detach().mean(dim=0).squeeze()
_std_mean = torch.sqrt(1. / prec_mean)
_sigma_chol_mean = torch.mm(torch.diag(_std_mean), corr_chol_mean)
sigma_mean = torch.mm(_sigma_chol_mean, _sigma_chol_mean.T)
return mu_mean, sigma_mean
def marginal(guide, num_samples=25):
posterior_predictive = Predictive(guide, num_samples=num_samples)
posterior_samples = posterior_predictive.forward(data)
mu_mean = posterior_samples['mu'].detach().mean(dim=0)
prec_mean = posterior_samples['prec'].detach().mean(dim=0)
corr_mean = torch.zeros(T, D, D)
for t in range(T):
corr_mean[t, ...] = posterior_samples['corr_chol_{}'.format(
t)].detach().mean(dim=0)
beta_mean = posterior_samples['beta'].detach().mean(dim=0)
weights_mean = mix_weights(beta_mean)
centers, sigmas, _ = truncate(alpha, mu_mean, prec_mean, corr_mean,
weights_mean)
return centers, sigmas
