Tutorials
Belief and Decision Networks

## Tutorial 2: Loading A Pre-existing Graph

The Bayes Applet comes with several pre-defined examples to allow you to start working with Bayesian networks without having to create one yourself. The applet has two main modes, Belief and Decision mode, each having similar but different options and functionality. The applet will change modes according to what type of sample network you are trying to load. To change modes manually, select either 'Belief Network Mode' or 'Decision Network Mode' under 'Network Options->Belief/Decision Modes'. To load an example file, go to the 'File' menu and select 'Load Sample Graph'. A dialog will open with a drop-down menu allowing you to select a particular example.

Belief Networks:

• Fire Alarm Belief Network: This example models the following problem: when somebody reports people leaving a building because a fire alarm went off, did it go off because of tampering or is there really a fire?
• Simple Diagnosic Example: This is example 6.14, shown in Figure 6.3 of Poole and Mackworth, Artificial Intelligence: foundations of computational agents.
• Simple Graph: This is an example of a singly connected network.
• Conditional Independence Quiz: The conditional independence quiz is not intended to be a network used for querying, but is a graph useful for thinking about conditional independence questions.
• Car Starting Problem: The car starting problem is a larger network intended to model the behaviour of a car, based on the state of its parts.
• Electrical Diagnosis Problem: The electrical diagnosis problem is from Computational Intelligence, Chapter 10.3, and models the wiring of a house.
• Diagnosis Problem: This example is from the ALARM monitoring system, a network to monitor intensive care patients:
I. Beinlich and G. Suermondt and R. Chavez and G. Cooper, The ALARM monitoring system: A case study with two probabilistic inference techniques for belief networks, Proc.~2'nd European Conf.~on AI and Medicine, Springer-Verlag, Berlin, 1989.
• Hailfinder Belief Net: A weather forecasting network built from:
Edwards, Ward (1998) "Hailfinder: Tools for and experiences with Bayesian Normative Modeling" in American Psychologist, 53, 416-428.

Decision Networks:

• Simple Umbrella Decision: This is a very simple decision problem built by Ross D. Shachter modeling the decision of whether or not to bring an umbrella based on the weather forecast.
• Fire Alarm Decision Problem: The fire alarm decision problem is similar to the Fire Alarm Belief Network, except that it models utilities and decisions as well as probabilities.
• Car Buying Decision: The car buying decision models the decision problem of a potential car buyer. The buyer is contemplating whether to buy a certain used car, and has the option of running certain tests on it beforehand. This network is built from:
Howard, Ronald A. (1983) "The used car buyer" in The Principles and Applications of Decision Analysis: Vol. II, Ronald A. Howard and J. E. Matheson (eds.), Strategic Decisions Group, Menlo Park, CA. Originally copyright 1962.
• Delivery Decision Problem: Decision network from Figure 9.5 of Poole and Mackworth, Artificial Intelligence: foundations of computational agents.