If you like the ai-junkie tutorials then you'll love this. Although it's aimed at game developers it's useful to anyone wanting to learn about genetic algorithms and artificial neural networks with as much fun and as few headaches as possible. Check out the reviews to see if it's the right book for you.
Reviews
Here are links to some reviews of the book.
Order
Table Of Contents
Foreword
From Steve Woodcock of Game AI.
Introduction from the author
An overview of the book and its aims
Introduction to Windows Programming
Chapter 1: Hello World!
A brief history of MS Windows.
Your first Windows program.
Creating your first window.
The Windows message pump
The Windows procedure
Keyboard input
Chapter 2: Further Adventures with Windows Programming
The windows GDI
Overview
Device contexts
Pens, Brushes, Lines, Curves and Colors
Shapes
Text
A real-time message pump
How to create and use a back buffer
Using Resources: Icons, Cursors, Menus, Dialog boxes
Timers
These first couple of chapters run you through the win32 programming stuff you need to know to understand the code projects in the book. If you already know this stuff then pretend it's not there! ;0)
Genetic Algorithms
Chapter 3: The Birds and the Bees
Evolution in nature
The digital approach
The genetic algorithm loop
Roulette wheel selection
Crossover
Mutation
Save Bob! - A Pathfinding Project
Bob’s environment
Encoding the chromosome
The Epoch function
Choosing the parameter values
Roulette wheel selection revisited
Crossover revisited
Mutation revisited
Running the program
Stuff to Try
Chapter 4: A Different Type of Encoding
Permutation encoding
The Travelling Salesman Problem
Traps to avoid
The permutation crossover operator
The exchange mutation operator
Deciding on a fitness function
Selection
Putting it all together
Stuff to try
Chapter 5: Our Salesman gets Smarter
Alternative permutation mutation operators
Alternative permutation crossover operators
Optimizing your GA
The fitness landscape
Selection Techniques
Elitism
Steady state
Stochastic Universal Sampling(SUS)
Tournament selection
Scaling Techniques
Rank scaling
Sigma scaling
Boltzmann scaling
Generic Crossover Operators
Single point
Two point
Multi-point
Niching Techniques
Fitness sharing
Stuff to try
Chapter 6: Learning to Land on the Moon
Introducing behavioral encoding
Description of the project
Mathematics and physics 101
The manned lunar lander project
The GA controlled lunar lander project
Description
Encoding the genome
The crossover and mutation operators
The fitness function
Running the program
Stuff to try
Neural Networks
Chapter 7: Neural Networks in Plain English
A Biological Neural Network – Your Brain
Structure and development of a human brain
Properties of a human brain
Artificial Neurons
The step activation function
Mathematical notation
Structure of artificial neural networks
Overview of supervised and unsupervised learning
The Smart Minesweeper Project
Choosing the outputs
The sigmoid activation function
Choosing the inputs
Deciding on hidden units
Descriptions of the classes used in the project
Encoding the neural networks.
A floating point genetic algorithm
Crossover
Mutation
Performance Improvements
A new crossover operator
Reducing the inputs
Stuff to try
Chapter 8: Giving your Bot Senses
Obstacle Avoidance
Creating sensors
A new fitness function
Project settings
Exploration
Creating a memory
Implementing the memory
Project settings
Recurrent networks
Summary
Stuff to try
Chapter 9: A Supervised Training Approach
Overview
The XOR Problem
Historical relevance
Linear separability
How does backpropagation work?
The equations
The learning rate
Example using the XOR problem
Changes to the CNeuralNet code
Mouse Gesture Recognition
Representing a gesture with vectors
Training the network
Recording and transforming the mouse data
Adding gestures
Tips and Tricks
The Momentum Rule
The error landscape
Local and global minima
Avoiding Overfitting
Reducing neurons
Adding jitter
Early stopping
The Softmax Activation Function
The cross-entropy error
A Modern Fable - the tank story
Applications
Stuff to try
Chapter 10: Evolving Neural Networks in Real-time – Brainy Aliens
Description of the Brainy Aliens project.
Overview of the technique used
Implementation
Roswell Revisited - An Alien Autopsy
Alien evolution
A different update method explained
Stuff to try
Chapter 11: Evolving Neural Net Topologies
Introduction
The competing conventions problem
Direct Encoding
GENITOR
Binary matrix
Node based
Path based
Indirect Encoding
Grammar based
Bi-dimensional growth encoding
N.E.A.T.
Overview
The Encoding method
Link genes
Node genes
The genome structure
Operators and Innovations
Historical markings and innovations
Adding a link
Adding a node
Designing a valid crossover operator
Mutation operators
Speciation
What is a species?
Compatibility testing
Fitness sharing
The NEAT epoch
Converting a Genome into a Phenotype
The neuron structure
The synapse structure
Putting it all together
Stuff to try
Final Words
Bibliography and recommended reading
Internet Resources