WHAT IS GENETIC ALGORITHM AND WHERE IS IT APPLIED

ABSTRACT

Let's figure out what a genetic algorithm is, how it works and where it can be applied.

Keywords: algorithm, genetic algorithm, programming.

First of all, let's find out what a genetic algorithm is. A genetic algorithm is an evolutionary model, an adapted process of natural evolution. At its core, the algorithm uses duplication operators and chromosome changes. The working principle of a genetic algorithm is based on the mechanisms of population genetics. Namely, random changes in the genotype, manipulation of the chromosome set and natural selection.

Let's look at how this algorithm works. We start by creating a primary population - a set of chromosomes. A chromosome is a set of values for gene inclusions. A chromosome is usually stored as a bit string, for example "11001011010". In the first generation, chromosomes are created randomly. To understand the fitness of a chromosome, there is a fitness function - it evaluates the fitness of a chromosome on a certain scale, for example from 0 to 100, where 0 is the least fit chromosome and 100 is the most successful.

Then the main part of the algorithm begins. The process of gene crossover takes place. There are several implementations, for example single-point crossover - two chromosomes are split in half, the new chromosome contains the first half of the first chromosome and the second half of the second chromosome. After new chromosomes have been produced, there is a certain probability that mutation will occur in several copies. Mutation is no less important a process because it allows us to obtain more diverse chromosomes.

All new chromosomes are evaluated according to their fitness function. The most successful are placed in a pantheon - a small repository of the best adapted chromosomes, which are passed down through generations until a replacement is found. This is an imitation of evolution - through random crossbreeding and random mutation, better chromosomes emerge!

This cycle continues until the required fitness threshold is reached or a certain number of iterations have been completed. As a result, we get a pretty good solution to the problem. Of course, there are many modifications to the algorithm, but this is for a deeper dive into the topic.

It remains to understand what the genetic algorithm can be used for. It is important to understand that the final result depends directly on the fitness function, which means that to implement the algorithm correctly, you need to understand the problem well and what you want to get in the end. For this reason, this algorithm cannot be used to solve complex problems. We can say that the genetic algorithm is an improved version of a simple enumeration.

It remains to understand what the genetic algorithm can be used for. It is important to understand that the final result depends directly on the fitness function, which means that to implement the algorithm correctly you need to understand the problem well and what you want to get in the end. For this reason, this algorithm cannot be used to solve complex problems. We can say that the genetic algorithm is an improved version of a simple enumeration.

Let's sum up. Genetic algorithms have limitations, but the benefits of using them are obvious. If you have a task to classify images, text, tabular data, or predict time series, you either experiment for a week or two, or you run Genetics and it builds an architecture overnight, and in a combination, you can't even imagine. Genetics is a powerful tool for typical tasks.

References:

1. Романов Д. Генетические алгоритмы: как это работает – 2023 – [электронный ресурс] – Режим доступа. – URL: https://www.comnews.ru/digital-economy/content/224864/2023-03-20/2023-w12/geneticheskie-algoritmy-kak-eto-rabotaet (дата обращения 13.12.2024)
2. Шишкова Н. А. Генетический алгоритм как метод оптимизации – 2017 – [электронный ресурс] – Режим доступа. – URL: https://cyberleninka.ru/article/n/geneticheskiy-algoritm-kak-metod-optimizatsii/viewer (дата обращения 13.12.2024)
3. Роль генетических алгоритмов в вопросах моделирования – 2022 – [электронный ресурс] – Режим доступа. – URL: https://habr.com/ru/articles/693742/ (дата обращения 13.12.2024)
4. Стариков А. Генетические алгоритмы — математический аппарат – 2021 – [электронный ресурс] – Режим доступа. – URL: https://loginom.ru/blog/ga-math (дата обращения 13.12.2024)