遗传算法基础：探索计算智能的进化策略

需积分: 9 78 浏览量更新于2024-07-17 1 收藏 2.37MB PDF 举报

《遗传算法精要》是Studies in Computational Intelligence系列的第679卷，由Oliver Kramer撰写。这本书深入探讨了遗传算法这一计算模型的基本原理与应用，它是基于达尔文自然选择和遗传学原理的一种优化方法。该算法模仿生物进化过程，通过随机变异、选择和交叉等操作，搜索并找到问题的潜在最优解，特别适用于解决复杂的优化问题，如组合优化、机器学习中的参数调整和规划问题。遗传算法的核心思想包括种群初始化、适应度函数定义、选择机制（如轮盘赌选择或锦标赛选择）、交叉（基因重组）、变异（随机改变个体特性）以及淘汰弱者等步骤。这些步骤共同构成了算法的迭代流程，使其能够在大量可能的解决方案中逐步逼近最佳解。在《遗传算法精要》中，作者不仅详细介绍了基本概念，还会涵盖高级主题，例如多目标优化、动态环境适应、以及并行和分布式遗传算法的应用。此外，书中还可能讨论了遗传算法与其他计算智能技术（如神经网络、模糊系统、进化计算等）的结合，展示了其广泛的应用领域，包括工程、计算机科学、物理学和生命科学等多个学科。由于该书的出版旨在快速传播和确保高质量，读者可以期待找到最新研究进展和设计方法，以及实例分析和实用案例，以便于理解和实践遗传算法。对于研究者来说，这是一本深入了解遗传算法理论和应用的宝贵参考资料；对于工程师和开发者而言，它提供了将这种优化技术应用于实际项目中的工具和策略。《遗传算法精要》是一本全面且深入的指南，不仅适合想要掌握遗传算法基础的人，也适合那些希望扩展优化工具箱的专业人士，帮助他们在解决复杂问题时，利用遗传算法的智慧来寻找最有效的解决方案。

8 1 Introduction

1.5 This Book

This book is an introduction to Genetic Algorithms. Genetic Algorithms

are biologically inspired optimization heuristics. Meanwhile established as solid

methodological playground, in particular for difﬁcult optimization problems, a long

line of research has proven their success, experimentally and theoretically. Conse-

quently, the list of topics associated with Genetic Algorithms has grown, which

is surely overwhelming for beginners in the ﬁeld. But also the expert may sometimes

be surprised about the variety of areas and evolving ﬁelds. Figure 1.4 visualizes a list

of typical keywords that are associated with Genetic Algorithms and that play

a role in the course of this book reaching from genetic operators like mutation and

selection to applications like learning and constraint handling. The variety of topics

is large and the list of research papers in each ﬁeld is overwhelming. However, when

having a closer look, many concepts are similar to each other, a baseline of similar

mechanisms and heuristics exists. It is the task of this book to identify the most

important ones and to compose them to a comprehensive depiction.

There are many good related depictions. Most of them put another focus on

the introductory concepts. Eiben and Smith [23] give a good overview to Genetic

Algorithms. The book is an attractive introduction of Genetic Algorithm

concepts and gives an overview to many related ﬁelds. A recommendation to all

researchers, who are interested in theoretical results, is the book by Neumann and

Witt [80]. It introduces basic proof techniques and exemplary shows i nteresting run-

time analysis results. An introduction to t he related ﬁeld of machine learning is

given by J ames et al. [42], which is an introductory variant of the famous counterpart

Elements of statistical learning [38].

This book is structured as follows.

• Chapter 2 gives an introduction to Genetic Algorithms concentrating on the

most important concepts like populations, the generational scheme, crossover,

mutation, selection, genotype-phenotype mapping, and termination conditions.

• Chapter 3 puts an emphasis on tuning and control of Genetic Algorithms’

parameters. It turns out that the genetic search crucially depends on the choice

of adequate parameters. The chapter gives an introduction to different tuning and

control strategies like dynamic control, Rechenberg’s mutation rate control, and

self-adaptation.

Fig. 1.4 Cloud of typical

keywords associated with

Genetic Algorithms

genetic algorithms

1/5th success rule

mutation rates

evolution strategies

multimodal

multi-objective

selection

mutation

recombination

1.5 This Book 9

• Chapter 4 present strategies for overcoming local optima in order to approximate

the global one, or at least to ﬁnd as many local optima as possible. Strategies

comprise restarts, niching, ﬁtness sharing, and novelty search.

• Chapter 5 gives an introduction to constraint handling. Practical optimization prob-

lems are often constrained. Not the whole solution space is allowed, but only a

feasible subset. Genetic Algorithms have to be adapted to cope with constraints.

One strategy is to choose representations and operators that avoid infeasible solu-

tions. An easy way to cope with constraints is the use of penalty functions, which

deteriorate the ﬁtness of a solution.

• Chapter 6 presents multi-objective optimization approaches. Multiple optimiza-

tion goals induce multiple objectives. If they are conﬂictive, the minimization

of one objective results in the maximization of another. Strategies like the non-

dominated sorting Genetic Algorithm, rake selection, and selection based on

the hypervolume indicator allow the approximation of a Pareto-set of solutions.

• Chapter 7 gives an introduction to theoretical research on Genetic Algorithms.

Starting with an overview of theoretical research in this area, a runtime analysis

is exemplarily presented and an overview of further theoretical tools is presented.

• Chapter 8 presents research in the intersection between machine learning and

Genetic Algorithms. Machine learning algorithms can be used to improve and

support genetic search. Examples include covariance matrix estimation, meta-

modeling, and visualization. Genetic Algorithms are effective tuning and learn-

ing approaches for machine learning problems.

• Chapter 9 shows Genetic Algorithms in various applications. Genetic Algo-

rithms can be used to optimize machine learning problems, to optimize wind tur-

bine locations considering wake effects and geo-constraints, to scale the features

of wind power data in nearest neighbor regression, and to optimize rule bases for

virtual power plants.

• Finally, Chap. 10 closes with a summary of the most important aspects and con-

tributions of this book.

The book closes with a summary and an appendix containing supplementary infor-

mation.

1.6 Further Remarks

This depiction passes on complex notations where possible. Although a thorough

mathematical formulation might be much more exact helping to implement and

model certain aspects and details, it also can complicate the understanding of con-

cepts, the context, and the connections between different mechanisms. However, here

is some basis notation that we usually employ in our depictions. A vector is a bold

small letter x while a scalar is written with a small plain letter.

Also the pseudocode of most algorithms presented in this book passes on com-

plex notations. More important than the interpretation of possibly arbitrary levels of

Chapter 2

Genetic Algorithms

2.1 Introduction

Genetic Algorithms are heuristic search approaches that are applicable to a wide

range of optimization problems. This ﬂexibility makes them attractive for many

optimization problems in practice. Evolution is the basis of Genetic Algorithms.

The current variety and success of species is a good reason for believing in the power

of evolution. Species are able to adapt to their environment. They have developed

to complex structures that allow the survival in different kinds of environments.

Mating and getting offspring to evolve belong to the main principles of the success

of evolution. These are good reasons for adapting evolutionary principles to solving

optimization problems.

In this chapter we will introduce the foundations of Genetic Algorithms.

Starting with an introduction to the basic Genetic Algorithm with populations, we

will introduce the most important genetic operators step by step, which are crossover,

mutation, and selection. Further, we will discuss genotype-phenotype mapping, com-

mon termination conditions, and give a short excursus to experimental analysis.

2.2 Basic Genetic Algorithm

The classic Genetic Algorithm is based on a set of candidate solutions that repre-

sent a solution to the optimization problem we want to solve. A solution is a potential

candidate for an optimum of the optimization problem. Its representation plays an

important role, as the representation determines the choice of the genetic operators.

Representations are usually lists of values and are more generally based on sets of

symbols. If they are continuous, they are called vectors, if they consist of bits, they

are called bit strings. I n case of combinatorial problems the solutions often consist of

symbols that appear in a list. An example is the representation of a tour in case of the

traveling salesman problem. Genetic operators produce new solutions in the chosen

O. Kramer, Genetic Algorithm Essentials, Studies in Computational

Intelligence 679, DOI 10.1007/978-3-319-52156-5_2

剩余93页未读，继续阅读

随机散步

粉丝: 0
资源: 10

遗传算法基础：探索计算智能的进化策略

The Implementation of Genetic Algorithm .pdf

A Genetic Algorithm Tutorial.pdf

genetic algorithm

cannot import name 'GeneticAlgorithm'

编写一段Python的遗传算法库genetic_algorithm中的GeneticAlgorithm类的带全部关键参数的示范

geneticalgorithm安装python

AttributeError: module 'pyeasyga.pyeasyga' has no attribute 'GeneticAlgorithm'

GeneticAlgorithm 是哪个库，怎么maven引入

'GeneticAlgorithm' object has no attribute 'run'

创建一个GeneticAlgorithm实例并配置关键参数，其中问题域为6输入5水平

最新资源