Practical Neural Network Recipies in C++

Contents

Preface χνϋ

1.

Foundations 1

Motivation 2

New Life for Old Techniques 3

Perceptrons and Linear Separability 4

Neural Network Capabilities 6

Basic Structure of a Neural Network 8

Training 9

Validation 10

Leave-/r-out Method 12

2.

Classification 15

Binary Decisions 16

Making the Decision 17

Multiple Classes 18

Reject Category 18

Making the Decision 19

Other Encoding Schemes 19

Supervised versus Unsupervised Training 21

3.

Autoassociation 23

Autoassociative Filtering 24

Code for Autoassociative Filtering 28

Noise Reduction 29

Learning a Prototype from Exemplars 31

Exposing Isolated Events 32

Pattern Completion 40

Error Correction 41