Other Evolutionary and Bio-Inspired Computation Techniques

In this chapter, you will broaden your horizons and discover several new problem-solving and optimization techniques related to genetic algorithms. Two different techniques of this extended family – genetic programming and particle swarm optimization – will be then demonstrated by implementing problem-solving Python programs. Finally, we will provide a brief overview of a number of other related computation paradigms.

This chapter will cover the following topics:

The evolutionary computation family of algorithms
Understanding the concepts of genetic programming and how they differ from genetic algorithms
Using genetic programming to solve the even parity check problem
Using particle swarm optimization to optimize Himmelblau's function
Understanding the principles behind several other evolutionary and...

Technical requirements

In this chapter, we will be using Python 3 with the following supporting libraries:

deap
numpy
networkx

The programs that will be used in this chapter can be found in this book's GitHub repository at https://github.com/PacktPublishing/Hands-On-Genetic-Algorithms-with-Python/tree/master/Chapter12.

Check out the following video to see the Code in Action:
http://bit.ly/2Gx4KsL

Evolutionary computation and bio-inspired computing

Throughout this book, we have covered the problem-solving technique known as genetic algorithms and applied it to numerous types of problems, including combinatorial optimization, constraint satisfaction, and continuous function optimization, as well as to machine learning and artificial intelligence. However, as we mentioned in Chapter 1, An Introduction to Genetic Algorithms, genetic algorithms are just one branch within a larger family of algorithms called evolutionary computation. This family consists of various related problem-solving and optimization techniques, all of which draw inspiration from Charles Darwin's theory of natural evolution.

The main characteristics that are shared by these techniques are as follows:

The starting point is an initial set (population) of candidate solutions.
The candidate solutions...

Genetic programming

Genetic programming (GP) is a special form of genetic algorithm – the technique we have been applying throughout this entire book. In this special case, the candidate solutions – or individuals – that we are evolving with the aim of finding the best one for our purpose are actual computer programs, hence the name. In other words, when we apply GP, we evolve computer programs with the goal of finding a program that will excel at performing a particular task.

As you may recall, genetic algorithms use a representation of the candidate solutions, often referred to as a chromosome. This representation is the subject of genetic operators, namely selection, crossover, and mutation. Applying these operators to the current generation results in a new generation of solutions that is expected to produce better results than its predecessor. In most...

Particle swarm optimization

Particle swarm optimization (PSO) draws its inspiration from natural groupings of individual organisms, such as flocks of birds or schools of fish, generally referred to as swarms. The organisms interact within the swarm without central supervision, working together toward a common goal. This observed behavior gave rise to a computational method that can solve or optimize a given problem by using a group of candidate solutions represented by particles analogous to organisms in a swarm. The particles move in the search space, looking for the best solution, and their movement is governed by simple rules that involve their position and velocity (directional speed).

The PSO algorithm is iterative, and in each iteration, every particle's position gets evaluated, and its best location so far, as well as the best location within the entire group of particles...

Other related techniques

Besides the techniques we have covered so far, there are numerous other problem-solving and optimization techniques that draw their inspiration from the Darwinian evolution theory, as well as from various biological systems and behaviors. The following subsections briefly describe several more of these techniques.

Evolution strategies

Evolution strategies (ES) are a kind of genetic algorithm that emphasizes mutation rather than crossover as the evolutionary facilitator. The mutation is adaptive, and its strength is learned over the generations. The selection operator in evolution strategy is always rank-based rather than done using actual fitness values. A simple version of this technique is called...

Summary

In this chapter, you were introduced to the extended family of evolutionary computation and some of the common characteristics of its members. Then, we used genetic programming – a special case of genetic algorithms – to implement the even parity check task. This was followed by creating a program that utilized the particle swarm optimization technique to optimize Himmelblau's function. We concluded this chapter with a brief overview of several other related problem-solving techniques.

Now that this book has come to its end, I wanted to thank you for taking this journey with me and going through the various aspects and use cases of genetic algorithms and evolutionary computation. I hope you found this book interesting as well as thought-provoking. As this book demonstrated, genetic algorithms and their related techniques can be applied to a plethora...