Select Index

Getting Started

    What Is Neural Networks?
    Using the Documentation
    Neural Network Applications
        Applications in this Toolbox
        Business Applications
        Summary

Examples

Neuron Model and Network Architectures

    Neuron Model
        Simple Neuron
        Transfer Functions
        Neuron with Vector Input

    Network Architectures
        A Layer of Neurons
        Multiple Layers of Neurons

    Data Structures
        Simulation with Concurrent Inputs in a Static Network
        Simulation with Sequential Inputs in a Dynamic Network
        Simulation with Concurrent Inputs in a Dynamic Network

    Training Styles
        Incremental Training (of Adaptive and Other Networks)
        Batch Training
        Training Tip

    Summary
        Equations

Perceptrons

    Introduction
        Important Perceptron Functions

    Neuron Model
    Perceptron Architecture
    Creating a Perceptron (newp)
        Simulation (sim)
        Initialization (init)

    Learning Rules
    Perceptron Learning Rule (learnp)
    Training (train)
    Limitations and Cautions
        Outliers and the Normalized Perceptron Rule

    Graphical User Interface
        Introduction to the GUI
        Create a Perceptron Network (nntool)
        Train the Perceptron
        Export Perceptron Results to Workspace
        Clear Network/Data Window
        Importing from the Command Line
        Save a Variable to a File and Load It Later

    Summary
        Equations
        New Functions

Linear Filters

    Introduction
    Neuron Model
    Network Architecture
        Creating a Linear Neuron (newlin)

    Mean Square Error
    Linear System Design (newlind)
    Linear Networks with Delays
        Tapped Delay Line
        Linear Filter

    LMS Algorithm (learnwh)
    Linear Classification (train)
    Limitations and Cautions
        Overdetermined Systems
        Underdetermined Systems
        Linearly Dependent Vectors
        Too Large a Learning Rate

    Summary
        Equations
        New Functions

Backpropagation

    Introduction
    Fundamentals
        Architecture
        Simulation (sim)
        Training

    Faster Training
        Variable Learning Rate (traingda, traingdx)
        Resilient Backpropagation (trainrp)
        Conjugate Gradient Algorithms
        Line Search Routines
        Quasi-Newton Algorithms
        Levenberg-Marquardt (trainlm)
        Reduced Memory Levenberg-Marquardt (trainlm)

    Speed and Memory Comparison
        Summary

    Improving Generalization
        Regularization
        Early Stopping
        Summary and Discussion

    Preprocessing and Postprocessing
        Min and Max (premnmx, postmnmx, tramnmx)
        Mean and Stand. Dev. (prestd, poststd, trastd)
        Principal Component Analysis (prepca, trapca)
        Post-Training Analysis (postreg)

    Sample Training Session
    Limitations and Cautions
    Summary

Control Systems

    Introduction
    NN Predictive Control
        System Identification
        Predictive Control
        Using the NN Predictive Controller Block

    NARMA-L2 (Feedback Linearization) Control
        Identification of the NARMA-L2 Model
        NARMA-L2 Controller
        Using the NARMA-L2 Controller Block

    Model Reference Control
        Using the Model Reference Controller Block

    Importing and Exporting
        Importing and Exporting Networks
        Importing and Exporting Training Data

    Summary

Radial Basis Networks

    Introduction
        Important Radial Basis Functions

    Radial Basis Functions
        Neuron Model
        Network Architecture
        Exact Design (newrbe)
        More Efficient Design (newrb)
        Demonstrations

    Generalized Regression Networks
        Network Architecture
        Design (newgrnn)

    Probabilistic Neural Networks
        Network Architecture
        Design (newpnn)

    Summary
        New Functions

Self-Organizing and Learn. Vector Quant. Nets

    Introduction
        Important Self-Organizing and LVQ Functions

    Competitive Learning
        Architecture
        Creating a Competitive Neural Network (newc)
        Kohonen Learning Rule (learnk)
        Bias Learning Rule (learncon)
        Training
        Graphical Example

    Self-Organizing Maps
        Topologies (gridtop, hextop, randtop)
        Distance Funct. (dist, linkdist, mandist, boxdist)
        Architecture
        Creating a Self Organizing MAP Neural Network (newsom)
        Training (learnsom)
        Examples

    Learning Vector Quantization Networks
        Architecture
        Creating an LVQ Network (newlvq)
        LVQ1 Learning Rule (learnlv1)
        Training
        Supplemental LVQ2.1 Learning Rule (learnlv2)

    Summary
        Self-Organizing Maps
        Learning Vector Quantization Networks
        New Functions

Adaptive Filters and Adaptive Training

    Introduction
        Important Adaptive Functions

    Linear Neuron Model
    Adaptive Linear Network Architecture
        Single ADALINE (newlin)

    Mean Square Error
    LMS Algorithm (learnwh)
    Adaptive Filtering (adapt)
        Tapped Delay Line
        Adaptive Filter
        Adaptive Filter Example
        Prediction Example
        Noise Cancellation Example
        Multiple Neuron Adaptive Filters

    Summary
        Equations
        New Functions

Applications

    Introduction
        Application Scripts

    Applin1: Linear Design
        Problem Definition
        Network Design
        Network Testing
        Thoughts and Conclusions

    Applin2: Adaptive Prediction
        Problem Definition
        Network Initialization
        Network Training
        Network Testing
        Thoughts and Conclusions

    Appelm1: Amplitude Detection
        Problem Definition
        Network Initialization
        Network Training
        Network Testing
        Network Generalization
        Improving Performance

    Appcr1: Character Recognition
        Problem Statement
        Neural Network
        System Performance
        Summary

Advanced Topics

    Custom Networks
        Custom Network
        Network Definition
        Network Behavior

    Additional Toolbox Functions
        Initialization Functions
        Transfer Functions
        Learning Functions

    Custom Functions
        Simulation Functions
        Initialization Functions
        Learning Functions
        Self-Organizing Map Functions

Historical Networks

    Introduction
        Important Recurrent Network Functions

    Elman Networks
        Architecture
        Creating an Elman Network (newelm)
        Training an Elman Network

    Hopfield Network
        Fundamentals
        Architecture
        Design (newhop)

    Summary
        New Functions

Network Object Reference

    Network Properties
        Architecture
        Subobject Structures
        Functions
        Parameters
        Weight and Bias Values
        Other

    Subobject Properties
        Inputs
        Layers
        Outputs
        Targets
        Biases
        Input Weights
        Layer Weights

Functions -- Categorical List

    Analysis Functions
    Distance Functions
    Graphical Interface Function
    Layer Initialization Functions
    Learning Functions
    Line Search Functions
    Net Input Derivative Functions
    Net Input Functions
    Network Functions
    Network Initialization Function
    Network Use Functions
    New Networks Functions
    Performance Derivative Functions
    Performance Functions
    Plotting Functions
    Pre- and Postprocessing Functions
    Simulink Support Function
    Topology Functions
    Training Functions
    Transfer Derivative Functions
    Transfer Functions
    Utility Functions
    Vector Functions
    Weight and Bias Initialization Functions
    Weight Derivative Functions
    Weight Functions

Transfer Function Graphs

Functions -- Alphabetical List

Glossary

Mathematical Notation

    Mathematical Notation for Equations and Figures
        Basic Concepts
        Language
        Weight Matrices
        Bias Elements and Vectors
        Time and Iteration
        Layer Notation
        Figure and Equation Examples

    Mathematics and Code Equivalents

Demonstrations and Applications

    Tables of Demonstrations and Applications
        Chapter 2: Neuron Model and Network Architectures
        Chapter 3: Perceptrons
        Chapter 4: Linear Filters
        Chapter 5: Backpropagation
        Chapter 7: Radial Basis Networks
        Chapter 8: Self-Organizing and Learn. Vector Quant. Nets
        Chapter 12: Recurrent Networks
        Chapter 9: Adaptive Networks
        Chapter 10: Applications

Simulink

    Block Set
        Transfer Function Blocks
        Net Input Blocks
        Weight Blocks

    Block Generation
        Example
        Exercises

Code Notes

    Dimensions
    Variables
        Utility Function Variables

    Functions
    Code Efficiency
    Argument Checking

Bibliography

Printable Documentation (PDF)

Product Page