## Three-Day Introductory Course in Washington, D.C.

Artificial Intelligence with Bayesian Networks and BayesiaLab

**University of Phoenix, 25 Massachusetts Ave NW, Classroom 105, Washington, D.C. 20001January 9–11, 2018, 9 a.m. to 5 p.m. each day**

Go beyond descriptive analytics and enter the realm of probabilistic and causal reasoning with Bayesian networks. Learn all about designing and machine-learning Bayesian networks with BayesiaLab (see complete course program)

This highly acclaimed course gives you a comprehensive introduction that allows you to employ Bayesian networks for applied research across many fields, such a biostatistics, decision science, econometrics, ecology, marketing science, sensory research, sociology, just to name a few.

The hallmark of this three-day course is that every segment on theory is immediately followed by a corresponding practice session using BayesiaLab. Thus, you have the opportunity to implement on your computer what the instructor just presented in his lecture. This includes knowledge modeling, probabilistic reasoning, causal inference, machine learning, probabilistic structural equation models, plus many more examples. Given the strictly limited class size, the instructor is always available to coach you one-on-one as you progress through the exercises.

After the end of the course, you can continue your studies as you will have access to a full 60-day license of BayesiaLab Professional. Additionally, two workbooks, plus numerous datasets and sample networks help you to experiment independently with Bayesian networks. To date, over 750 researchers from all over the world have taken this course (see testimonials). For most of them, Bayesian networks and BayesiaLab have become crucial tools in their research.

## Day 1: Theoretical Introduction

## Introduction

- Bayesian Networks: Artificial Intelligence for Decision Support under Uncertainty
- Probabilistic Expert System
- A Map of Analytic Modeling and Reasoning
- Bayesian Networks and Cognitive Science
- Unstructured and Structured Particles Describing the Domain
- Expert Based Modeling and/or Machine Learning
- Predictive vs. Explanatory Models, i.e., Association vs. Causation
- Application Examples: Medical Expert Systems, Stock Market Analysis, Microarray Analysis, Consumer Segmentation, Drivers Analysis, and Product Optimization

## Examples of Probabilistic Reasoning

- Cognitive Science: How Our Probabilistic Brain Uses Priors in the Interpretation of Images
- Interpreting Results of Medical Tests
- Kahneman & Tversky’s Yellow Cab/White Cab Example
- The Monty Hall Problem, Solving a Vexing Puzzle with a Bayesian Network
- Simpson’s Paradox - Observational Inference vs Causal Inference

## Probability Theory

- Probabilistic Axioms
- Interpretation with Particles
- Joint Probability Distribution (JPD)
- Probabilistic Expert System for Decision Support: Types of Requests
- Leveraging Independence Properties
- Product/Chain Rule for Compact Representation of JPD

## Bayesian Networks

- Qualitative Part: Directed Acyclic Graph
- Graph Terminology
- Graphical Properties
- D-Separation
- Markov Blanket
- Quantitative Part: Marginal and Conditional Probability Distributions
- Exact and Approximate Inference in Bayesian networks
- Example of Probabilistic Inference: Alarm System

## Building Bayesian Networks Manually

- Expert-Based Modeling via Brainstorming
- Why Expert-Based Modeling?
- Value of Expert-Based Modeling
- Structural Modeling: Bottom-Up and Top-Down Approaches
- Parametric Modeling
- Cognitive Biases
- BEKEE: Bayesia Expert Knowledge Elicitation Environment

## Day 2: Machine Learning—Part 1

## Parameter Estimation

- Maximum Likelihood Estimation
- Bayesian Parameter Estimation with Dirichlet Priors
- Smooth Probability Estimation (Laplacian Correction)

## Information Theory

- Information is a Measurable Quantity: Log-Loss
- Expected Log-Loss
- Entropy
- Conditional Entropy
- Mutual Information
- Symmetric Relative Mutual Information
- Kullback-Leibler Divergence

## Unsupervised Structural Learning

- Entropy Optimization
- Minimum Description Length (MDL) Score
- Structural Coefficient
- Minimum Size of Data Set
- Search Spaces
- Search Strategies
- Learning Algorithms
- Maximum Weight Spanning Tree
- Taboo Search
- EQ
- TabooEQ
- SopLEQ
- Taboo Order

- Data Perturbation
- Example: Exploring the relationships in Body Dimensions
- Data Import (Typing, Discretization)
- Definition of Classes
- Exclusion of a Node
- Heuristic Search Algorithms
- Data Perturbation (Learning, Bootstrap)
- Choice of the Structural Coefficient
- Console
- Symmetric Layout
- Analysis of the Model (Arc Force, Node Force, Pearson Coefficient)
- Dictionary of Node Positions
- Adding a Background Image

## Supervised Learning

- Learning Algorithms
- Naive
- Augmented Naive
- Manual Augmented Naive
- Tree-Augmented Naive
- Sons & Spouses
- Markov Blanket
- Augmented Markov Blanket
- Minimal Augmented Markov Blanket

- Variable Selection with Markov Blanket
- Example: Predictions based on body dimensions
- Data Import (Data Type, Supervised Discretization)
- Heuristic Search Algorithms
- Target Evaluation (In-Sample, Out-of-Sample: K-Fold, Test Set)
- Smoothed Probability Estimation
- Analysis of the Model (Monitors, Mapping, Target Report, Target Posterior Probabilities, Target Interpretation Tree)
- Evidence Scenario File
- Automatic Evidence-Setting
- Adaptive Questionnaire
- Batch Labeling

## Day 3: Machine Learning—Part 2

## Semi-Supervised Learning—Variable Clustering

- Algorithms
- Example: S&P 500 Analysis
- Variable Clustering
- Changing the number of Clusters
- Dynamic Dendrogram
- Dynamic Mapping
- Manual Modification of Clusters
- Manual Creation of Clusters

- Semi-Supervised Learning
- Search Tool (Nodes, Arcs, Monitors, Actions)

- Variable Clustering

## Data Clustering

- Synthesis of a Latent Variable
- Expectation-Maximization Algorithm
- Ordered Numerical Values
- Cluster Purity
- Cluster Mapping
- Log-Loss and Entropy of the Data
- Contingency Table Fit
- Hypercube Cells per State
- Example: Segmentation of men based on body dimensions
- Data Clustering (Equal Frequency Discretization, Meta-Clustering)
- Quality Metrics (Purity, Log-Loss, Contingency Table Fit)
- Posterior Mean Analysis (Mean, Delta-Means, Radar Charts)
- Mapping
- Cluster Interpretation with Target Dynamic Profile
- Cluster Interpretation with Target Optimization Tree
- Projection of the Cluster on Other Variables

## Probabilistic Structural Equation Models

- PSEM Workflow
- Unsupervised Structural Learning
- Variable Clustering
- Multiple Clustering for Creating a Factor Variable (via Data Clustering) per Cluster of Manifest Variables
- Unsupervised Learning for Representing the Relationships Between the Factors and the Target Variables

- Example: The French Market of Perfumes
- Cross-Validation of the Clusters of Variables
- Displayed Classes
- Total Effects
- Direct Effects
- Direct Effect Contributions
- Tornado Analysis
- Taboo, EQ, TabooEQ, and Arc Constraints
- Multi-Quadrants
- Export Variations
- Target Optimization with Dynamic Profile
- Target Optimization with Tree

## About the Instructor

Dr. Lionel Jouffe is co-founder and CEO of France-based Bayesia S.A.S. Lionel holds a Ph.D. in Computer Science from the University of Rennes and has been working in the field of Artificial Intelligence since the early 1990s. While working as a Professor/Researcher at ESIEA, Lionel started exploring the potential of Bayesian networks. After co-founding Bayesia in 2001, he and his team have been working full-time on the development BayesiaLab, which has since emerged as the leading software package for knowledge discovery, data mining and knowledge modeling using Bayesian networks. BayesiaLab enjoys broad acceptance in academic communities as well as in business and industry.

## Who should attend?

Applied researchers, statisticians, data scientists, data miners, decision scientists, biologists, ecologists, environmental scientists, epidemiologists, predictive modelers, econometricians, economists, market researchers, knowledge managers, marketing scientists, operations researchers, social scientists, students and teachers in related fields.

## What's required?

- Basic data manipulation skills, e.g., with Excel.
- No prior knowledge of Bayesian networks is required.
- No programming skills are required. You will use the graphical user interface of BayesiaLab for all exercises.

For a general overview of this field of study, we suggest that you download a free copy of our book, Bayesian Networks & BayesiaLab. Although by no means mandatory, reading its first three chapters would be an excellent preparation for the course.

## Terms & Conditions

- You may cancel your registration for a full refund of the course fees up to 30 days before the start of the course. If you cancel within 30 days of the event, your course fee will not be refunded. However, you will be able to apply 100% of the paid course fees towards future BayesiaLab courses.
- Accommodation is at the participants' own expense.

## Testimonials from Earlier Courses

"I would absolutely recommend this course as a thorough and in-depth introduction to Bayesian Networks and the BayesiaLab package. The small class sizes also contributed to an enjoyable and engaging learning experience."**—Brian Potter, Infotools (Introductory Course in Melbourne, November 2015)**.

"This is one of the best **—Vladimir Agajanov, Moody's (Introductory Course in New York, January 2016)**.

"Overall, this training was outstanding. Lionel is a gifted teacher, and it helps that you are showcasing a **—Felix Elwert, PhD, Vilas Associate Professor of Sociology, University of Wisconsin-Madison (Introductory Course in Chicago, May 2014)**.

“A must-take course for anyone looking to leverage advanced Bayesian network techniques in virtually any domain.”**—****Alex Cosmas, Chief Scientist, Booz Allen Hamilton ( Introductory Course in Los Angeles, June 2011)**.

“The BayesiaLab software is impressive in its sophistication and multi-faceted abilities as a decision support tool. I had been using it primarily as a modeling tool for deductive analysis. Taking this class opened my eyes to BayesiaLab's incredible data-mining abilities. If you are looking for something that will provide a totally new angle on business decision problems, this is it!”**—****Michael Ryall, PhD, Professor of Strategy and Economics, Rotman Business School, University of Toronto (Introductory Course in Chicago, July 2013)**.

"This class can only be described as eye-opening, the tool as terrific. Some of the best instruction for the shortest period of time I’ve ever received. A seriously terrific job.” **—****Beau Martin, President of American Choice Modeling (Introductory Course in Chicago, July 2013)**.