Get Started with SamIam

Welcome to the SamIam program! If you are a new user of SamIam, and you have access to a Windows computer, the first thing we recommend you to do is to view the introductory video tutorial (WMV / MP4) - it gives a basic introduction to the program, including: the differences between Edit Mode and Query Mode, how to select nodes, how to view posterior probabilities, how to set evidence, the evidence tree, and some information about supported file formats. After you have gotten familiar with the basics of the program, you will probably want to view the following tutorials that we have geared toward some of the more important tools that come with SamIam:

Creating Networks (WMV / MP4) - covers the functions SamIam provides to edit the structure of Bayesian networks, including: adding and deleting variable nodes and adding and deleting dependencies between variables (edges).

Inference (WMV / MP4) - introduces the four algorithms SamIam furnishes and explains what types of queries each algorithm supports. Also, introduces the new Evidence Impact tool.

Sensitivity Analysis (WMV / MP4) - illustrates every feature of the Sensitivity Analysis tool through three natural examples, including: one-variable and two-variable constraints, unsatisfiable constraints, single-parameter and multiple-parameter suggestions, and how to automatically adopt suggested changes.

Time-Space Tradeoffs (WMV / MP4) - an example using the recursive conditioning algorithm. Provides guidance on how to best exploit the any-space features, and how to calculate Pr(e) using the least possible memory.

MAP/MPE (WMV / MP4) - shows how to use both the Maximum a Posteriori (MAP) tool and and Most Probable Explanation (MPE) tool, and uses an intuitive example as the basis for a discussion of the differences between MAP and MPE.

EM Learning (WMV / MPEG4) - shows how to use the SamIam EM Learning tool to learn CPT parameters from a Hugin format case file, and how to generate simulated case file data.

If you would like to use SamIam to work with belief networks created using the Genie tool, we hope that you find useful the Genie Files video tutorial (WMV / MP4) - an in-depth discussion of what features of Genie files SamIam does and does not support, including: diagnosis node types, submodels, noisy or semantics, and some information on trouble-shooting SamIam's Genie file support. Also, please refer to Working with Files.

We deliver SamIam with invocation scripts that pass necessary arguments to the Java virtual machine: "samiam.bat" on Windows and "runsamiam" on Solaris/Linux. In particular, these files pass the argument '-Xmx' to Java in order to increase the maximum size of the runtime memory allocation pool to 512 megabytes. You may choose to edit your run script in order to increase or decrease the memory available to Java. Java’s default memory allocation, 64 megabytes, is insufficient to compile larger, more complex networks.