Leveraging Domain Knowledge in Multitask Bayesian Network Structure Learning

Network structure learning algorithms have aided network discovery in fields such as bioinformatics, neuroscience, ecology and social science. However, challenges remain informative networks for related sets of tasks because the search space Bayesian structures is characterized by large basins approximately equivalent solutions. Multitask select a set that are near each other space, rather than score-equivalent chosen from independent regions space. This selection preference allows domain expert to see only differences supported data. usefulness these scientific datasets limited existing naively assume all pairs equally related. We introduce framework relaxes this assumption incorporating knowledge about task-relatedness into objective. Using our framework, we first multitask algorithm leverages relatedness tasks. use explore effect on show learns closer ground truth naive discovers patterns interesting.