Extensions
What follows in an incomplete and brief list of features that are currently being implemented.
- At the moment it necessary that the world is fixed, e.g. the relational structure is given. We are currently implementing a novel method for making inference in domains with reference uncertainty. This is unpublished work.
- An expectation maximization (EM) algorithm for missing data and latent variables.
- Support for continuos variables
- Parametric distributions (e.g. Normal, Beta distributions)
- Universal prior distributions in conjunction with parametric distributions