Résumé
A neural network is a computational network composed of
mathematically defined elements that are thought to
approximate the workings of biological neurons. In
situations where a decision depends on many factors, neural
networks provided decisions with more consistency and
accuracy than trained a human. Networks are based on
substantial theoretical foundations and have great
practical utility in many situations. Any problem that has
traditionally been solved using statistical methods or
through traditional modeling can most likely be more
effectively solved using neural networks. The field of
neural networks has developed to the point where the
underlying code can be formalized, realized as objects, and
used in countless applications. Therefore, the powerful
concepts of object-oriented design and object-oriented
programming can now be implemented fully in creating neural
network architectures. Instead of focusing on the mechanics
of the underlying foundation, neural network objects allow
the designer to focus upon the enhancement of an existing
architecture or the design of a new one. This book is the
first to take full advantage of the reusable nature of
classes in neural network design.
Subject Index.
Object-Oriented Neural Networks in C++ is a valuable tool for anyone who wants to understand, implement, or utilize neural networks. This book/disk package provides the reader with a foundation from which any neural network architecture can be constructed. The author has employed object-oriented design and object-oriented programming concepts to develop a set of foundation neural network classes, and shows how these classes can be used to implement a variety of neural network architectures with a great deal of ease and flexibility. A wealth of neural network formulas (with standardized notation), object code implementations, and examples are provided to demonstrate the object-oriented approach to neural network architectures and to facilitate the development of new neural network architectures. This is the first book to take full advantage of the reusable nature of neural network classes.
KEY FEATURES
- Describes how to use the classes provided to implement a variety of neural network architectures including ADALINE, Backpropagation, Self-Organizing, and BAM
- Provides a set of reusable neural network classes, created in C++, capable of implementing any neural network architecture
- Includes an IBM disk of the source code for the classes, which is platform independent
Table of contents
Preface.Introduction: Biological Roots.
- Types of Neural Networks.
Software Engineering and Object-Oriented Programming.
Other Attempts to Construct Object-Oriented Neural Networks.
Overview of the Remaining Chapters.
- Inheritance.
Virtual Functions.
Polymorphism and Dynamic Binding.
Template Classes.
- Pattern Set Formalization.
Formalized Neural-Network Parameters.
- Realizing the ADALINE Neural Network Objects.
ADALINE_Link Class.
ADALINE Training Set.
Creating the ADALINE Network.
Training the ADALINE Network.
ADALINE_Network Class.
ADALINE Neural-Network Objects Summary.
- Realizing the Backpropagation Neural Network
Objects.
Using the Backpropagation Neural-Network Objects.
Backpropagation Neural Network Node.
Backprop Application.
Epoch Training with Backpropagation Neural Networks.
Generic Backpropagation Neural Network Class.
Backpropagation Neural Network Examples.
Summary of Backpropagation Neural-Network Objects.
- Realizing the Self-Organizing Neural-Network
Objects.
Using the SON Objects.
SON_Network Class.
SON Application.
SON Examples.
Summary on SON Objects.
- Realizing the BAM Objects.
Using the BAM Objects.
BAM Network Node.
BAM System.
BAM Objects Summary.
Subject Index.
Caractéristiques techniques
| PAPIER | |
| Éditeur(s) | Apress |
| Auteur(s) | Joey Rogers |
| Parution | 10/10/1996 |
| Nb. de pages | 310 |
| EAN13 | 9780125931151 |
| ISBN13 | 978-0-12-593115-1 |
Avantages Eyrolles.com
Livraison à partir de 0,01 € en France métropolitaine
Paiement en ligne SÉCURISÉ
Livraison dans le monde
Retour sous 15 jours
+ d'un million et demi de livres disponibles
Nos clients ont également acheté
Consultez aussi
- Les meilleures ventes en Graphisme & Photo
- Les meilleures ventes en Informatique
- Les meilleures ventes en Construction
- Les meilleures ventes en Entreprise & Droit
- Les meilleures ventes en Sciences
- Les meilleures ventes en Littérature
- Les meilleures ventes en Arts & Loisirs
- Les meilleures ventes en Vie pratique
- Les meilleures ventes en Voyage et Tourisme
- Les meilleures ventes en BD et Jeunesse
- Informatique Développement d'applications Technologies objet
- Informatique Développement d'applications Programmation graphique et multimédia Programmation graphique en C/C++
- Informatique Développement d'applications Langages C++
- Informatique Développement d'applications Outils de développement Borland C++ Builder
- Informatique Développement d'applications Outils de développement Visual C++
- Informatique Développement d'applications Outils de développement Visual C++ .NET
