pict

HUGIN Training - Module Overview

The Hugin training course covers Bayesian Networks, how to build knowledge bases using Hugins graphical user interface, how to analyze results, and programming using the Hugin Decision Engine. It contains a large number of examples, exercises, and hands-on experiences. The exercises are solved using a trial version of Hugin Developer.

Module Objectives

When the trainee completes this module s/he will have acquired the following knowledge, methodologies and capabilities:

  • Fundamental understanding of Bayesian networks and limited memory influence diagrams (LIMIDs).
  • Fundamental understanding of the modelling process, including object-oriented modelling.
  • Methodology and templates for model development, including tips and tricks for improving model efficiency.
  • Fundamental understanding of the processes and methods for analysing the results produced by a model.
  • Capability to apply the Hugin Graphical User Interface for efficient model development.
  • Capability to program with the Hugin Decision Engine for application development.

HUGIN Training Process

The 3-days Hugin course gives each participant a fundamental introduction to Bayesian networks and LIMIDs through a combination of theory and practice. The theory is limited to supporting the intuitive explanation of different features of Bayesian network and LIMID methodologies.

The course is organized into a set of sessions. Each session considers a particular topic, which is explained in detail using examples. For instance, one topic is the introduction of Bayesian networks while another topic is modelling techniques for Bayesian networks. During each session, students will have the opportunity to solve hands-on exercises on their own PC. To solve exercises, each student will have access to the Hugin Developer tool.

Unless otherwise specified the training starts at 9 o'clock and ends at 16 o'clock.

The topics considered in the course are as follows:

Day 1 - Bayesian Networks

Course Introduction
Graph & Probability theory
Bayesian Networks
Programming w. HUGIN API
Modelling Techniques I

Day 2 - Influence Diagrams

Modelling Techniques I
Learning Bayesian networks
Decision and Utility Theory
Influence Diagrams & LIMIDs
Modelling Techniques II

Day 3 - Methods of Analysis

Object-Oriented Networks
Data Conflict Analysis
Value of Information Analysis
Evidence Sensitivity Analysis
Parameter Sensitivity Analysis

Material

The course is based on the book Bayesian Networks and Influence Diagrams - A Guide to Construction and Analysis by Uffe Kjærulff and Anders L Madsen. Students do not need to read this book, to participate in the HUGIN course.

The course material includes a CD-ROM with course slides and a trial-license to HUGIN Developer. Each student will also receive a set of hardcopy slides.

Requirements

The two main prerequisites of the course are limited knowledge of how to use a computer and basic mathematical skills.

All lessons will be in English, and our instructor has more than 10 years of comprehensive professional training experience.

The course requires the participants to bring their own laptop. For participants unable to bring their own laptop, computer rentals are $100 per day and must be reserved in advance.

Instructor

Anders L. Madsen

Anders holds a PhD in Decision Support Systems. Anders has more than 10 years of teaching experience.

 anders