Parameter Learning of Bayesian Network with Multiplicative Synergistic Constraints

Learning the conditional probability table (CPT) parameters of Bayesian networks (BNs) is a key challenge in real-world decision support applications, especially when there are limited data available. The traditional approach to this introducing domain knowledge/expert judgments that encoded as qualitative parameter constraints. In paper, we focus on multiplicative synergistic negative synergy constraint and positive paper symmetric. order integrate constraints into learning process Network parameters, propose four methods deal with based idea classical isotonic regression algorithm. simulated by using lawn moist model Asia network, compared them maximum likelihood estimation (MLE) Simulation results show proposed superior MLE algorithm accuracy learning, which can improve obtain more accurate estimators parameters. reduce dependence expert experiences. Combining these under small sample conditions.