Neural Network Programming with Java By Alan Souza, Fábio Soares PDF

Neural Network Programming with Java By Alan Souza, Fábio Soares PDF

Informations about the book:

Title: Neural Network Programming with Java

Author: Lentin Joseph

Size: 5.8

Format: EPUB

Year: 2016

Pages: 244

Book Contents:

Chapter 1: Getting Started with Neural Networks 

Discovering neural networks 

Why artificial neural network? 

How neural networks are arranged 

The very basic element – artificial neuron 

Giving life to neurons – activation function 

The fundamental values – weights 

An important parameter – bias 

The parts forming the whole – layers 

Learning about neural network architectures 

Monolayer networks 

Multilayer networks 

Feedforward networks 

Feedback networks 

From ignorance to knowledge – learning process 

Let the implementations begin! Neural networks in practice 

Summary 

Chapter 2: How Neural Networks Learn 

Learning ability in neural networks 

How learning helps to solve problems 

Learning paradigms 

Supervised learning 

Unsupervised learning 

Systematic structuring – learning algorithm 

Two stages of learning – training and testing 

The details – learning parameters 

Error measurement and cost function 

Table of Contents

[ ii ]

Examples of learning algorithms 

Perceptron 

Delta rule 

Coding of the neural network learning 

Learning parameter implementation 

Learning procedure 

Class definitions 

Two practical examples 

Perceptron (warning system) 

ADALINE (traffic forecast) 

Summary 

Chapter 3: Handling Perceptrons 

Studying the perceptron neural network 

Applications and limitations of perceptrons 

Linear separation 

Classical XOR case 

Popular multilayer perceptrons (MLPs) 

MLP properties 

MLP weights 

Recurrent MLP

MLP structure in an OOP paradigm 

Interesting MLP applications 

Classification in MLPs 

Regression in MLPs 

Learning process in MLPs 

Simple and very powerful learning algorithm – Backpropagation 

Elaborate and potent learning algorithm – Levenberg–Marquardt 

Hands-on MLP implementation! 

Backpropagation in action 

Exploring the code 

Levenberg–Marquardt implementation 

Practical application – types of university enrolments 

Summary 

Chapter 4: Self-Organizing Maps 

Neural networks' unsupervised way of learning 

Some unsupervised learning algorithms 

Competitive learning or winner takes all 

Table of Contents

[ iii ]

Kohonen self-organizing maps (SOMs) 

One-Dimensional SOM 

Two-Dimensional SOM 

Step-by-step of SOM learning 

How to use SOMs 

Coding of the Kohonen algorithm 

Exploring the Kohonen class 

Kohonen implementation (clustering animals) 

Summary 

Chapter 5: Forecasting Weather 

Neural networks for prediction problems 

No data, no neural net – selecting data 

Knowing the problem – weather variables 

Choosing input and output variables 

Removing insignificant behaviors – Data filtering 

Adjusting values – data preprocessing 

Equalizing data – normalization 

Java implementation for weather prediction 

Plotting charts 

Handling data files 

Building a neural network for weather prediction

Empirical design of neural networks 

Choosing training and test datasets 

Designing experiments 

Results and simulations 

Summary 

Chapter 6: Classifying Disease Diagnosis 117

What are classification problems, and how can neural 

be applied to them? 

A special type of activation function – Logistic regression 

Multiple classes versus binary classes 

Comparing the expected versus produced results – the

confusion matrix 

Classification measures – sensitivity and specificity 

Applying neural networks for classification 

Disease diagnosis with neural networks 

Using ANN to diagnose breast cancer 

Applying NN for an early diagnosis of diabetes 

Summary 

Table of Contents

[ iv ]

Chapter 7: Clustering Customer Profiles 

Clustering task 

Cluster analysis 

Cluster evaluation and validation 

External validation 

Applied unsupervised learning 

Neural network of radial basis functions 

Kohonen neural network 

Types of data 

Customer profiling 

Preprocessing data 

Implementation in Java 

Card credit analysis for customer profiling 

Summary 

Chapter 8: Pattern Recognition (OCR Case) 

What is pattern recognition all about? 

Definition of classes among tons of data 

What if the undefined classes are undefined? 

External validation 

How to apply neural networks in pattern recognition 

Preprocessing the data 

The OCR problem 

Simplifying the task – digit recognition 

Approach to digit representation 

Let the coding begin! 

Generating data 

Building the neural network 

Testing and redesigning – trial and error 

Results 

Summary 

Chapter 9: Neural Network Optimization and Adaptation 

Common issues in neural network implementations 

Input selection 

Data correlation 

Dimensionality reduction 

Data filtering 

Structure selection

Table of Contents

[ v ]

Online retraining

Stochastic online learning 

Implementation 

Application 

Adaptive neural networks

Adaptive resonance theory 

Implementation 

Summary 

Appendix A: Setting up the NetBeans Environment 

Download and install NetBeans 

Setting up the NetBeans environment 

Importing a project 

Programming and running code with NetBeans 

Debugging with NetBeans 

Appendix B: Setting Up the Eclipse Environment 

Download and install Eclipse 

Setting up the Eclipse environment 

Importing a project 

Programming and running code with the Eclipse IDE 

Debugging with the Eclipse IDE 

Appendix C: References 

Chapter 1 – Getting Started with Neural Networks 

Chapter 2 – How Neural Networks Learn 

Chapter 3 – Working with Perceptrons 

Chapter 4 – Self-Organizing Maps 

Chapter 5 – Forecasting the Weather 

Chapter 6 – Disease Diagnosis 

Chapter 7 – Clustering Customer Profiles 

Chapter 8 – Pattern Recognition (the OCR Case) 

Chapter 9 – Neural Network Optimization and Adaptation