Skip to main content


Fig. 7 | EURASIP Journal on Bioinformatics and Systems Biology

Fig. 7

From: Using multi-step proposal distribution for improved MCMC convergence in Bayesian network structure learning

Fig. 7

Sampled posteriors using different proposal distributions. For each different one-step and two-step combination shown in the figure, 100 chains were run with random initial DAGs for a burn-in period of 1,000,000 and a sample of 100,000 was taken. Each position on the x-axis represents a different DAG, sorted by ascending score, which is shown with a dashed line. The x-axes are the same between subfigures so a DAG at position k is the same in all subfigures. The posterior probabilities of each DAG in the samples are shown with dots (zero values not shown)

Back to article page