Learning Bayesian Networks

Bayesian Network (BN) is an intuitive, graphical representation of a joint probability distribution of a set of random variables with a possible mutual causal relationship. BN is a machine-learning data analysis technique and BN modelling approach complements the traditional statistical data analysis approach in overcoming the curse of dimensionality and naturally capturing the independence and dependence relationships among model variables.

Netica is the most widely used BN software in the world. If you can use Microsoft Word and Excel confidently in your work, you will be able to learn Bayesian Networks using Netica. It is the conditional probability and hierarchical structure matter in modelling a complex real life problem and BN is one of the most suitable and powerful tools to do it. This 90-minute presentation will (intends to) cover the following topics:

  • Concepts and definitions about what is a Bayesian Network (BN); a very brief introduction to the theoretical basis of a BN model.
  • Demonstration of various BN applications through 10 examples (from the simplest toy example to some examples which are based on real life data sets) using Netica.
  • An overview introduction to a 3-day workshop on Learning BNs by Hands-on Practice with Examples.

Fee-based workshop

A 3-day workshop (Wagga campus) on learning and application of Bayesian Network models (Bayes net and decision net models, and Dynamic Bayesian Network models) is available now: the workshop is designed/developed for maximizing the hands-on experiences in learning BNs from examples and keep the theoretical preliminary requirements / theoretical exposition to the minimum. The goal is to establish participants’ ability to build BN models for solving the problems from their research or professional activities.

Learning outcomes

Learning Bayesian Networks by hands-on practice with examples.

By the end of the first two days, participants:

  1. should be able to  build simple BN models (Bayes net and decision net models) using the Netica BN software package
  2. are able to run simple analysis given a BN model
  3. are able to use Netica example models for improving his/her knowledge about the application of BN models
  4. have a good idea what BN models can do and the strength and limitations of BN model applications.

By the end of the third day, participants

  1. should be able to  build non-trivial BN models (Bayes net and decision net models) and simple Dynamic Bayesian Network (DBN) models in solving their research / real life problems
  2. participants are able to learn a BN model (determining model structure and estimating CPTs) and testing the model performance directly from a data file
  3. are able to perform a reasonable analysis given a BN model and understand the strength and limitation of the analysis outcomes
  4. participants are confident to use Netica example models for improving his/her skills and/or knowledge about the application of BN models.

Cost

Wagga campus

  • CSU students and staff, or DPI staff: $300 (including GST) per person.

The cost includes morning tea, a light lunch, and afternoon tea.

Other campuses

For logistics reasons, the workshop run on other campuses will be slightly modified to a 2.5-day workshop.

  • CSU students and staff, or DPI staff: $200 (including GST) per person.

The cost includes morning tea, a light lunch, and afternoon tea will be provided for the first two days. Morning tea only for the third day.

Expressions of interest

Please send us your Expression of Interest: