GitHub - pengzhaostat/SMF-structured-variational-inference (original) (raw)
Reproducing examples
simu_mcmc_binary_loop :
reprducing results for the comparison between SMF, MCMC and MF for binary dynamic latent space model (Figure 2,3)
simu_mcmc_Gaussian_loop:
reprducing results for the comparison between SMF, MCMC and MF for Gaussian dynamic latent space model (Table 2)
classroom :
real data example, visualization of evolution of the estimated latent positions for the classroom dataset (Figure 5,6,7,8)
enron
real data example, missing link predictions for the enron dataset (Figure 4, 14)
simu_nodewise
simulation example to show the adavantage of using a nodewise adaptive priors (Figure 9)
simu_binary_innerproduct
simulation exmaple of recovering the inner products (Figure 10)
Gaussian_simu_hyper_IG and Gaussian_simu_hyper
sensitivity analysis of hyperparameters (Figure 11,12,13)
Simple simulation examples
simu_binary_simple :
simulation example, comparison between SMF and MF for binary dynamic latent space model
simu_Gaussian_simple:
simulation example, comparison between SMF and MF for Gaussian dynamic latent space model
simu_binary_mcmc :
simulation example, comparison between SMF and MCMC for binary dynamic latent space model
simu_Gaussian_mcmc:
simulation example, comparison between SMF and MCMC for Gaussian dynamic latent space model
List of functions
mix_DN_adaptive :
function, SMF for binary dynamic latent space model
mix_Gaussian_adaptive :
function, SMF for Gaussian dynamic latent space model
mean_field_DN_adaptive :
function, mean-field for binary dynamic latent space model
mean_field_Gaussian_adaptive :
function, mean-field for Gaussian dynamic latent space model
mcmc_DN_adaptive :
function, mcmc for binary dynamic latent space model
mcmc_Gaussian_adaptive :
function, mcmc for Gaussian dynamic latent space model
helper :
some helpful functions
mix_DN_adaptive_node
function, SMF for binary dynamic latent space model with nodewise adaptivity
mix_DN_adaptive_miss mix_DN_adaptive_invgamma_miss
function, SMF for binary dynamic latent space model with missing links