Table of Contents

1. Introduction

• Features of This Package

2. Neural Network Theory--A Short Tutorial

• Introduction to Neural Networks
  • Function Approximation
  • Time Series and Dynamic Systems
  • Classification and Clustering
• Data Preprocessing
• Linear Models
• The Perceptron
• Feedforward and Radial Basis Function Networks
  • Feedforward Neural Networks
  • Radial Basis Function Networks
  • Training Feedforward and Radial Basis Function Networks
• Dynamic Neural Networks
• Hopfield Network
• Unsupervised and Vector Quantization (VQ) Networks
• Further Reading

3. Getting Started and Basic Examples

• Palettes and Loading the Package
  • Palettes
  • Loading the Package and Data
• Package Conventions
  • Data Format
  • Function Names
  • Network Format
• NetClassificationPlot
• Basic Examples
  • Classification Problem Example
  • Function Approximation Example

4. The Perceptron

• Perceptron Network Functions and Options
  • InitializePerceptron
  • PerceptronFit
  • NetInformation
  • NetPlot
• Examples
  • Two Classes in Two Dimensions
  • Several Classes in Two Dimensions
  • Higher-Dimensional Classification
• Further Reading

5. The Feedforward Neural Network

• Feedforward Network Functions and Options
  • InitializeFeedForwardNet
  • NeuralFit
  • NetInformation
  • NetPlot
  • LinearizeNet and NeuronDelete
  • SetNeuralD, NeuralD, and NNModelInfo
• Examples
  • Function Approximation in One Dimension
  • Function Approximation from One to Two Dimensions
  • Function Approximation in Two Dimensions
• Classification with Feedforward Networks
• Further Reading

6. The Radial Basis Function (RBF) Network

• RBF Network Functions and Options
  • InitializeRBFNet
  • NeuralFit
  • NetInformation
  • NetPlot
  • LinearizeNet and NeuronDelete
  • SetNeuralD, NeuralD, and NNModelInfo
• Examples
  • Function Approximation in One Dimension
  • Function Approximation from One to Two Dimensions
  • Function Approximation in Two Dimensions
• Classification with RBF Networks
• Further Reading

7. Training Feedforward and Radial Basis Function Networks

• NeuralFit
• Examples of Different Training Algorithms
• Train with FindMinimum
• Troubleshooting
• Regularization and Stopped Search
  • Regularization
  • Stopped Search
  • Example
• Separable Training
  • Small Example
  • Larger Example
• Options Controlling Training Results Presentation
• The Training Record
• Writing Your Own Training Algorithms
• Further Reading

8. Dynamic Neural Networks

• Dynamic Network Functions and Options
  • Initializing and Training Dynamic Neural Networks
  • NetInformation
  • Predicting and Simulating
  • Linearizing a Nonlinear Model
  • NetPlot--Evaluate Model and Training
  • MakeRegressor
• Examples
  • Identifying the Dynamics of a DC Motor
  • Identifying the Dynamics of a Hydraulic Actuator
  • Bias-Variance Tradeoff--Avoiding Overfitting
  • Fix Some Parameters--More Advanced Model Structures
• Further Reading

9. Hopfield Networks

• Hopfield Network Functions and Options
  • HopfieldFit
  • NetInformation
  • HopfieldEnergy
  • NetPlot
• Examples
  • Discrete-Time Two-Dimensional Example
  • Discrete-Time Classification of Letters
  • Continuous-Time Two-Dimensional Example
  • Continuous-Time Classification of Letters
• Further Reading

10. Unsupervised Networks

• Unsupervised Network Functions and Options
  • InitializeUnsupervisedNet
  • UnsupervisedNetFit
  • NetInformation
  • UnsupervisedNetDistance, UnUsedNeurons, and NeuronDelete
  • NetPlot
• Examples without SOM
  • Clustering in Two-Dimensional Space
  • Clustering in Three-Dimensional Space
  • Pitfalls with Skewed Data Density and Badly Scaled Data
• Examples with SOM
  • Mapping from Two to One Dimension
  • Mapping from Two Dimensions to a Ring
  • Adding a SOM to an Existing Unsupervised Network
  • Mapping from Two to Two Dimensions
  • Mapping from Three to One Dimension
  • Mapping from Three to Two Dimensions
• Change Step Length and Neighbor Influence
• Further Reading

11. Vector Quantization

• Vector Quantization Network Functions and Options
  • InitializeVQ
  • VQFit
  • NetInformation
  • VQDistance, VQPerformance, UnUsedNeurons, and NeuronDelete
  • NetPlot
• Examples
  • VQ in Two-Dimensional Space
  • VQ in Three-Dimensional Space
  • Overlapping Classes
  • Skewed Data Densities and Badly Scaled Data
  • Too Few Codebook Vectors
• Change Step Length
• Further Reading

12. Application Examples

• Classification of Paper Quality
  • VQ Network
  • RBF Network
  • Feedforward Network
• Prediction of Currency Exchange Rate

13. Changing the Neural Network Structure

• Change the Parameter Values of an Existing Network
  • Feedforward Network
  • RBF Network
  • Unsupervised Network
  • Vector Quantization Network
• Fixed Parameters
• Select Your Own Neuron Function
  • The Basis Function in an RBF Network
  • The Neuron Function in a Feedforward Network
• Accessing the Values of the Neurons
  • The Neurons of a Feedforward Network
  • The Basis Functions of an RBF Network