Tutorial on Learning Bayesian Networks for Complex Relational Data
Presenters: Oliver Schulte and Ted Kirkpatrick
Duration: 4 hours (including 30 min break)
Intended Audience: Researchers with a background in machine learning who wish to apply machine learning to relational data, by combining graphical probabilistic models with first-order logic.
Code and Datasets
- Sample databases are available in various formats. IMDb movie data serve as the running example.
- System Code is linked here.
Slides
The supplementary slides provide further details but will not be discussed during the tutorial.
- Section 0: Introduction.
- Section 1: Relational Data: Definition
- Section 2: First-order Bayesian Networks
- Section 3: Parameter Learning
- Section 4: Structure Learning
- Section 5: Relational Classification and Dependency Networks
- Section 6: Relational Outlier Detection and Exception Mining
- Supplementary Material: Definitions. Concise summary of the mathematical content. No examples.
Sample Bayesian Networks
The Bayesian networks can be opened using the AIspace tool.
