Manual for the implementation of neural networks in MATLAB

Table of Contents

1. Introduction

2. Neural Networks

2.1. What are neural networks?

2.2 Biological background

2.3 General structure of neural networks

2.4 Properties of neural networks

2.5 Learning in neural networks and weighting

2.6 Historical overview

2.7 Different networks models

2.7.1 Single-layer feed-forward networks

2.7.2 Multi-layer feed-forward networks

2.7.3 Recurrent networks

3. MATLAB

3.1 General overview

3.2. MATLAB

3.3 Matrixes in MATLAB

4. Realization of neural networks in MATLAB

4.1 Creation of neural networks with the Network/Data Manager

4.2 Import of data

4.3 The implementation of an ADALINE network

4.4 The implementation of a back-propagation network

4.5 Self-optimizing neural networks

4.6 Hopfield network

4.7 Summary

5. Conclusion

Objectives and Research Themes

The primary objective of this bachelor thesis is to provide a practical manual for implementing various types of artificial neural networks using the MATLAB software environment, serving as a supplement for a Master-level course.

• Theoretical foundations of biological and artificial neural networks.
• Fundamentals of the MATLAB software and its matrix manipulation capabilities.
• Step-by-step guidance for implementing network models via the Network/Data Manager.
• Practical examples covering ADALINE, back-propagation, self-optimizing, and Hopfield networks.

Excerpt from the Book

Chapter Summaries

1. Introduction: Presents the thesis as a manual for implementing neural networks in MATLAB, outlining the scope and the decision to focus on practical application rather than complex mathematical proofs.

2. Neural Networks: Provides a comprehensive overview of biological foundations, the general structure of artificial networks, historical milestones, and classifications of common neural network models.

3. MATLAB: Explains the basic functions of the MATLAB environment, specifically focusing on matrix manipulation, interface usage, and commands relevant to supporting neural network implementations.

4. Realization of neural networks in MATLAB: Details the practical implementation of several neural network architectures using the MATLAB Network/Data Manager, including data import procedures and specific training examples.

5. Conclusion: Summarizes the thesis, highlighting that while MATLAB is a powerful tool for initial neural network implementation, deeper complexities may require further advanced programming beyond the scope of the manual.

Keywords

Neural networks, MATLAB, Artificial Intelligence, Back-propagation, ADALINE, Hopfield network, Self-organizing maps, Neural Network Toolbox, Pattern recognition, Learning algorithms, Data implementation, Matrix manipulation, Supervised learning, Unsupervised learning, Network topology.

Frequently Asked Questions

What is the core focus of this thesis?

The thesis focuses on providing a practical guide for implementing artificial neural networks within the MATLAB software environment for educational purposes.

What are the primary thematic areas covered?

The work covers neural network theory, MATLAB fundamentals, network structure classification, and practical implementation procedures using the software's graphical user interface.

What is the main objective or research question?

The objective is to document how various neural network models can be successfully created and trained using the "Neural Network Toolbox" in MATLAB.

Which scientific methodology is utilized?

The author uses an experimental and descriptive approach, moving from theoretical basics to step-by-step practical implementation of specific network types.

What topics are discussed in the main part?

The main part covers the setup of the MATLAB environment, importing training data, configuring different network architectures, and evaluating training performance through graphical output.

Which keywords characterize this work?

Core keywords include Neural Networks, MATLAB, Artificial Intelligence, Back-propagation, ADALINE, and Hopfield Networks.

How is the "Network/Data Manager" used in this thesis?

It is used as the primary interface for creating, initializing, and training neural networks, avoiding the need for deep manual coding of every function.

What distinction does the author make between ADALINE and back-propagation networks?

The author distinguishes them based on their ability to solve linear versus nonlinear problems, noting that back-propagation is required for more complex tasks like the XOR function.

Why was the Hopfield network included as an example?

It serves as an example of a dynamic, recursive network that operates differently from feed-forward models, highlighting the versatility of neural architecture.