EDML: A Method for Learning Parameters in Bayesian Networks
This work addresses parameter estimation in Bayesian networks for researchers in machine learning, but it appears incremental as it builds on existing methods like EM without claiming major breakthroughs.
The paper tackles the problem of learning MAP parameters in binary Bayesian networks with incomplete data, proposing the EDML method which assumes Beta priors and can also learn maximum likelihood parameters with uninformative priors, showing interesting behaviors compared to EM through analytical and empirical studies.
We propose a method called EDML for learning MAP parameters in binary Bayesian networks under incomplete data. The method assumes Beta priors and can be used to learn maximum likelihood parameters when the priors are uninformative. EDML exhibits interesting behaviors, especially when compared to EM. We introduce EDML, explain its origin, and study some of its properties both analytically and empirically.