Link Search Menu Expand Document
CS310 - Machine Intelligence

Lab6 - Bayesian belief networks (BBN), (5pts)

In this lab you will learn to model and use a small BBN using the Hugin Lite software. Create a folder Lab6BBN and save your work there.

  1. Watch this 8min video by Anders Madsen to learn how to create and use a BBN.

  2. Install Hugin Lite from here.

  3. Open the Hugin Lite program.

  4. Use your browser to access the Hugin RESOURCES page which has many cases, demos, articles, tutorials etc. Under Tutorials, click the Building a BN Tutorial.

  5. Complete the Tutorial 1. Building a Bayesian Network, selecting folder Lab6BBN whenever you are told to save files. (Note: sometimes the tutorial does not match the GUI interface, but use your intuition. For ex., when naming/creating states in a node, use the “Add” bar to add states instead of “+-“ buttons; to see/edit the CPT table of a node, right click the node an select “Open Tables”). Be sure to perform a save after completing the last step of the tutorial.

  6. In a MS word file (or .txt file) titled Lab6BBN.docx and saved in your Lab6BBN, give a formula for calculating, P(Sick="Sick",Dry="Not"| Loses="no"), use the Bayesian net you have created to obtain the two necessary values, and show the final calculations.

  7. Repeat the previous step to compute P(Sick="No",Dry="Dry"| Loses="no").

Note: to resert you BBN, click the button with the curved arrow (“Initialize network”).

Grading: Submit your folder Lab6BBN containing the docx file and Apple.hkb file at the provided link.