Applying Neural Networks to Business, Industry, and Government

This course focuses on the theory and application of neural networks to modeling and classification problems in medical, insurance, credit, process, and scientific environments.

Syllabus

Day 1 Introduction to Neural Networks

- Historical Background
- Biological Inspirations
- Graphical Formulas
- Comparison to Other Technologies
- Terminology
- Key Characteristics
Lab: Exploring the Mechanisms of Back-Propagation


Day 2 Non-Linear Feed-Forward Neural Networks

- Network Training
- Back-Propagation
- Extended Delta Bar Delta
- Quick Prop
- Logicon Projection Network
- Modular Neural Networks
- Stochastic Networks
Lab: Predicting Low Birth Weights


Day 3 Neural Network Development

- Data Collection
- Train, Test, and Validation sets
- Data Analysis and Transformation
- Issues in Data Analysis and Transformation
- Data Types
- Symbolic or Enumerated Fields
- Numeric Enumerated Fields
- Continuous Numeric Data
- Data Transformations
- Continuous Encoding
- Binary Encoding
Lab: Predicting Risks for Insurance Underwriting


Day 4 Neural Network Development

- Data Transformations (continued)
- One-of-N Transformations
- Symbolic One-of-N
- Fuzzy One-of-N
- Variable Selection
- Statistical Approaches
- Correlated Inputs
- Neural Network Approach
- Genetic Selection
Labs: Classifying Glass from Its Physical Characteristics
         X-Squared


Day 5 Neural Network Development

- Model Development and Optimization
- Architecture
- Diagnostic Tools