Conference PaperPDF Available

An Overview of Evolutionary Computation

  • April 1993
DOI:10.1007/3-540-56602-3_163
  • Conference: Machine Learning: ECML-93, European Conference on Machine Learning, Vienna, Austria, April 5-7, 1993, Proceedings
Authors:
William Spears
William Spears
Kenneth De Jong at George Mason University
Kenneth De Jong
Thomas Bäck at Leiden University
Thomas Bäck
David B. Fogel at Trials.ai
David B. Fogel
  • Trials.ai
Show all 5 authorsHide
Download full-text PDF
Read full-text
Download citation

Abstract

Evolutionary computation uses computational models of evolutionary processes as key elements in the design and implementation of computer-based problem solving systems. In this paper we provide an overview of evolutionary computation, and describe several evolutionary algorithms that are currently of interest. Important similarities and differences are noted, which lead to a discussion of important issues that need to be resolved, and items for future research.
Content uploaded by Thomas Bäck
Author content
All content in this area was uploaded by Thomas Bäck on Sep 03, 2014
Content may be subject to copyright.
post
match Rule ListMessage List
input
message output
message
Output interface
effectors
Bucket brigade
(adjusts rule strengths)
(generates new rules)
Genetic algorithm
Payoff
Environment
Input interface
detectors
... Evolutionary computation solves problems by mimicking evolution steps in nature [6] [7]. Mimicking a natural process gives some abilities [8][9] to evolutionary computation. ...
View
... Evolutionary computation is a set of techniques that employ principles of natural evolution, such as reproduction, mutation and selection to solve problems and optimize systems [79]. In the context of video game AI, evolutionary computation can be utilized to enhance the performance and capabilities of AI characters within a game. ...
Artificial intelligence for video game visualization, advancements, benefits and challenges
Article
Full-text available
  • Jul 2023
In recent years, the field of artificial intelligence (AI) has witnessed remarkable progress and its applications have extended to the realm of video games. The incorporation of AI in video games enhances visual experiences, optimizes gameplay and fosters more realistic and immersive environments. In this review paper, we systematically explore the diverse applications of AI in video game visualization, encompassing machine learning algorithms for character animation, terrain generation and lighting effects following the PRISMA guidelines as our review methodology. Furthermore, we discuss the benefits, challenges and ethical implications associated with AI in video game visualization as well as the potential future trends. We anticipate that the future of AI in video gaming will feature increasingly sophisticated and realistic AI models, heightened utilization of machine learning and greater integration with other emerging technologies leading to more engaging and personalized gaming experiences.
View
... Additionally, it may be ineffective in MORL problems with continuous action-state spaces, where exhaustive sampling is challenging. The basic idea of MOEAs [33] comes from Darwin's natural selection mechanism of survival of the fittest [34]- [36]. Its purpose is to make individuals better adapt to the surrounding environment. ...
Efficient Elitist Cooperative Evolutionary Algorithm for Multi-Objective Reinforcement Learning
Article
Full-text available
  • Jan 2023
Sequential decision-making problems with multiple objectives are known as multi-objective reinforcement learning . In these scenarios, decision-makers require a complete Pareto front that consists of Pareto optimal solutions. Such a front enables decision-makers to understand the relationship between objectives and make informed decisions from a broad range of solutions. However, existing methods may be unable to search for solutions in concave regions of the Pareto front or lack global optimization ability, leading to incomplete Pareto fronts. To address this issue, we propose an efficient elitist cooperative evolutionary algorithm that maintains both an evolving population and an elite archive. The elite archive uses cooperative operations with various genetic operators to guide the evolving population, resulting in efficient searches for Pareto optimal solutions. The experimental results on submarine treasure hunting benchmarks demonstrate the effectiveness of the proposed method in solving various multi-objective reinforcement learning problems and providing decision-makers with a set of trade-off solutions between travel time and treasure amount, enabling them to make flexible and informed decisions based on their preferences. Therefore, the proposed method has the potential to be a useful tool for implementing real-world applications.
View
... Evolutionary computation methods arose from taking inspiration from biological mechanisms to design and implement computer-based problem-solving systems (Spears et al., 1993). This collection of methods allowed the creation of evolving and adaptive solutions to complex problems, especially the ones that impose challenges to traditional algorithms, such as randomness, chaotic disturbances, and complex non-linear dynamics, as outlined by Fogel (2000). ...
Analyzing the Impact of Strategic Behavior in an Evolutionary Learning Model Using a Genetic Algorithm
Article
Full-text available
This study presents an experimental approach to strategic behavior and economic learning by integrating game theory and Genetic Algorithms in a novel heuristic-based simulation model. Inspired by strategic scenarios that change over time, we propose a model where games can change based on agents’ behavior. The goal is to document the model design and examine how strategic behavior impacts the evolution of optimal outcomes in various choice scenarios. For diversity, 144 unique 2×2 games and three different strategy selection criteria were used: Nash equilibrium, Hurwicz rule, and a random selection technique. The originality of this study is that the introduced evolutionary algorithm changes the games based on their overall outcome rather than changing the strategies or player-specific traits. The results indicated optimal player scenarios for both The Nash equilibrium and Hurwicz rules, the first being the best-performing strategy. The random selection method failed to converge to optimal values in most of the selected populations, acting as a control feature and reinforcing the need for strategic behavior in evolutionary learning. Two further observations were recorded. First, games were frequently transformed so agents could coordinate their strategies to create stable optimal equilibria. Second, we observed the evolution of game populations into groups of fewer (repeating) isomorphic games with strong preceding game characteristics.
View
... Although crossover may be disruptive and split apart a good individual in addition to being helpful in bringing two useful components together, the selection pressure in GAs is often constructive. Since disruption of good solutions is practically guaranteed when they are close to the best solution, mimetic algorithms are frequently utilized for the closing stages of such situations where crossover can be very disruptive [26], [36] and [61]. The offspring will be organically adapted to the environment, much like native species, if crossovers occur and the more successful parents are selected more frequently than the less successful ones. ...
New Gradient Optimization and Genetic Algorithms Hybridized for Fast Convergence
Thesis
  • Dec 2022
In this thesis, three new search directions are proposed to improve the convergence speed of the steepest descent (SD) method for solving systems of linear equations where their matrices are symmetric and positive definite. These methods are motivated from the previous work introduced by Zubai’ah-Mustafa-Rivaie-Ismail (ZMRI) to modified the SD method [2]. Theoretical prove of descent condition for each presented method is given. Experimental results in tables and figures confirm that the three new methods are superior to SD and ZMRI methods with respect to the number of iterations NOI and the elapsed CPU time. Moreover, it is suggested that the hybridization of the proposed new algorithms with genetic algorithm is to reach the global solution with fast convergence. Numerical experiments show that the proposed hybrid methods are faster than the methods without hybrid and the standard GA in all cases. Furthermore, the results are more encouraging when the dimension and the condition number of the matrices are increased.
View
A systematic framework for evolutionary multi-objective optimization to complex building design problems at the early design stage
Article
Full-text available
  • Oct 2023
As in biology, multi-objective evolutionary algorithms cross-reference various design parameters in the search to maximize or minimize one or more specific objectives, thus finding the best solution for the specified purposes. In addition, it is possible to perform the appendment of many variables simultaneously and make numerous real-time simulations. This paper proposes a systematic framework for evolutionary multi-objective optimization to complex building design problems at the early stage. The framework is demonstrated by optimizing the courtyard geometry as a case study. The methods include generating courtyard geometry (i.e., height/width ratios and orientations) as design variables according to solar geometry. Simulations are explored, providing recommendations to maximize solar access in winter and filled shade during summer. The outcomes are a framework resumed systematically to address the contrasting objectives of the given building problems. The framework’s application can adapt to each case’s architectural, environmental, and technical criteria.
View
MOEA/D vs. NSGA-II: A Comprehensive Comparison for Multi/Many Objective Analog/RF Circuit Optimization through a Generic Benchmark
Article
  • Sep 2023
Thanks to the enhanced computational capacity of modern computers, even sophisticated analog/RF circuit sizing problems can be solved via electronic design automation (EDA) tools. Recently, several analog/RF circuit optimization algorithms have been successfully applied to automatize the analog/RF circuit design process. Conventionally, metaheuristic algorithms are widely used in optimization process. Among various nature-inspired algorithms, evolutionary algorithms (EAs) have been more preferred due to their superiorities (robustness, efficiency, accuracy etc.) over the other algorithms. Furthermore, EAs have been diversified and several distinguished analog/RF circuit optimization approaches for single-, multi-, and many- objective problems have been reported in the literature. However, there are conflicting claims on the performance of these algorithms and no objective performance comparison has been revealed yet. In the previous work, only a few case study circuits have been under test to demonstrate the superiority of the utilized algorithm, so a limited comparison has been made for only these specific circuits. The underlying reason is that the literature lacks a generic benchmark for analog/RF circuit sizing problem. To address these issues, we propose a comprehensive comparison of the most popular two evolutionary computation algorithms, namely Non-Sorting Genetic Algorithm-II (NSGA-II) and Multi-Objective Evolutionary Algorithm based Decomposition (MOEA/D), in this paper. For that purpose, we introduce two ad-hoc testbenches for analog (ANLG) and radio frequency (RF) circuits including the common building blocks. The comparison has been made at both multi- and many- objective domains and the performances of algorithms have been quantitatively revealed through the well-known Pareto-optimal front quality metrics.
View
Evolutionary algorithms for a simheuristic optimization of the product-service system design
Article
Full-text available
Offering Product-Service Systems (PSS) becomes an established strategy for companies to increase the provided customer value and ensure their competitiveness. Designing PSS business models, however, remains a major challenge. One reason for this is the fact that PSS business models are characterized by a long-term nature. Decisions made in the development phase must take into account possible scenarios in the operational phase. Risks must already be anticipated in this phase and mitigated with appropriate measures. Another reason for the design phase being a major challenge is the size of the solution space for a possible business model. Developers are faced with a multitude of possible business models and have the challenge of selecting the best one. In this article, a simheuristic optimization approach is developed to test and evaluate PSS business models in the design phase in order to select the best business model configuration beforehand. For optimization, a proprietary evolutionary algorithm is developed and tested. The results validate the suitability of the approach for the design phase and the quality of the algorithm for achieving good results. This could even be transferred to already established PSS.
View
View
View
View
View
View
Simulation of Genetic Systems by Automatic Digital Computers I. Introduction
Article
Methods of setting automatic digital computers to simulate the algebraic aspects of reproduction, segregation, and selection are discussed. The application of these methods to the problem of the importance of linkage in multifactorial inheritance is illustrated by results from the SILLIAC.
View
View
View
View
View
Evolutionary Operation: A Method for Increasing Industrial Productivity
Article
  • Jun 1957
The rate at which industrial processes are improved is limited by the present shortage of technical personnel. Dr Box describes a method of process improvement which supplements the more orthodox studies and is run in the normal course of production by plant personnel themselves. The basic philosophy is introduced that industrial processes should be run so as to generate not only product, but also information on how the product can be improved.
View
Dynamic Neighbourhood Size in Simulated Annealing
Article
  • Jan 1992
Simulated annealing has been shown to be a powerful stochastic method of tackling hard combinatorial optimisation problems, but it demands a vast amount of computa- tion time to arrive at a good approximate solution. A lot of research has been done on the cooling schedule of simulated annealing to speed up its convergence, but only limited attention has been paid to the impact of the neighbourhood size on the performance of simulated annealing. It has been shown that the performance of simulated annealing can be improved by adopting a suitable neighbourhood size. However, previous studies usually assumed that the neighbourhood size was fixed during search after decided at the beginning. This paper presents a simulated annealing algorithm with a dynamic neighbourhood size which depends on the current "temperature" value during search. A method of dynamically deciding the neighbourhood size by approximating a continu- ous probability distribution is given. Four continuous probability distributions are used in our experiments to generate neighbourhood sizes dynamically, and the results are compared.
View