SMILE Tutorials: Introduction

From DSL

This section reviews informally the architecture of SMILE from the point of view of an application programmer. We will do this by describing an example of a C/C++ program using SMILE. The program illustrates the most important functions of the library. Each of the objects used are described in more detail in the SMILE Reference Manual section. The complete code of the tutorials is included in the appendices.

We start by showing how to create a Bayesian network in a C++ program. This type of "manual" construction can be tedious it is much easier to do it from another program specialized in model building, for instance, Decision Systems Laboratory's Graphical Network Interface (GeNIe). This program is available for download from GeNIe Web site . In the second tutorial, we show how to perform inference in a Bayesian network. We describe how to introduce and remove evidence, how to invoke Bayesian updating algorithms, and how to check the results of the inference. The third tutorial transforms the Bayesian network into an influence diagram by adding one decision node and one value node. The fourth tutorial shows how to make inference on an influence diagram, how to decide which is the best policy, and how to interpret the results. Finally, the fifth tutorial shows how to compute the expected value of information.