1,721,280 research outputs found
Optimising the wine supply chain
No full textM. Michalewicz, Z. Michalewicz and R. Spittyhttp://www.awitc.com.au
Variants of Evolutionary Algorithms for Real-World Applications
No full textEvolutionary Algorithms (EAs) are population-based, stochastic search algorithms that mimic natural evolution. Due to their ability to find excellent solutions for conventionally hard and dynamic problems within acceptable time, EAs have attracted interest from many researchers and practitioners in recent years. This book “Variants of Evolutionary Algorithms for Real-World Applications” aims to promote the practitioner’s view on EAs by providing a comprehensive discussion of how EAs can be adapted to the requirements of various applications in the real-world domains. It comprises 14 chapters, including an introductory chapter re-visiting the fundamental question of what an EA is and other chapters addressing a range of real-world problems such as production process planning, inventory system and supply chain network optimisation, task-based jobs assignment, planning for CNC-based work piece construction, mechanical/ship design tasks that involve runtime-intense simulations, data mining for the prediction of soil properties, automated tissue classification for MRI images, and database query optimisation, among others. These chapters demonstrate how different types of problems can be successfully solved using variants of EAs and how the solution approaches are constructed, in a way that can be understood and reproduced with little prior knowledge on optimisation
Forecasting with a dynamic window of time: The DyFor genetic program model
No full textThe original publication is available at www.springerlink.comSeveral studies have applied genetic programming (GP) to the task of forecasting with favourable results. However, these studies, like those applying other techniques, have assumed a static environment, making them unsuitable for many real-world time series which are generated by varying processes. This study investigates the development of a new "dynamic" GP model that is specifically tailored for forecasting in non-static environments. This Dynamic Forecasting Genetic Program (DyFor GP) model incorporates methods to adapt to changing environments automatically as well as retain knowledge learned from previously encountered environments. The DyFor GP model is realised and tested for forecasting efficacy on real-world economic time series, namely the U.S. Gross Domestic Product and Consumer Price Index Inflation. Results show that the DyFor GP model outperforms benchmark models from leading studies for both experiments. These findings affirm the DyFor GP's potential as an adaptive, non-linear model for real-world forecasting applications and suggest further investigations
An evolutionary approach to practical constraints in scheduling: a case-study of the wine bottling problem
No full textPractical constraints associated with real-world problems are a key differentiator with respect to more artificially formulated problems. They create challenging variations on what might otherwise be considered as straightforward optimization problems from an evolutionary computation perspective. Through solving various commercial and industrial problems using evolutionary algorithms, we have gathered experience in dealing with practical dynamic constraints. Here, we present proven methods for dealing with these issues for scheduling problems. For use in real-world situations, an evolutionary algorithm must be designed to drive a software application that needs to be robust enough to deal with practical constraints in order to meet the demands and expectations of everyday use by domain specialists who are not necessarily optimization experts. In such situations, addressing these issues becomes critical to success. We show how these challenges can be dealt with by making adjustments to genotypic representation, phenotypic decoding, or the evaluation function itself. The ideas presented in this chapter are exemplified by the means of a case study of a real-world commercial problem, namely that of bottling wine in a mass-production environment. The methods described have the benefit of having been proven by a full-fledged implementation into a software application that undergoes continual and vigorous use in a live environment in which time-varying constraints, arising in multiple different combinations, are a routine occurrence.Arvind Mohais, Sven Schellenberg, Maksud Ibrahimov, Neal Wagner, and Zbigniew Michalewic
A Fuzzy-Evolutionary Approach to the Problem of Optimisation and Decision-Support in Supply Chain Networks
No full textThis chapter deals with the problem of balancing and optimising the multi-echelon supply chain network of an Australian ASX Top 50 company which specialises in the area of manufacturing agricultural chemicals. It takes into account sourcing of raw material, the processing of material, and the distribution of the final product. The difficulty of meeting order demand and balancing the plants’ utilisation while adhering to capacity constraints is addressed as well as the distribution and transportation of the intermediate and final products. The aim of the presented system is to minimise the time it takes to generate a factory plan while providing better accuracy and visibility of the material flow within the supply chain. The generation of factory plans within a short period of time allows for what-if-scenario analysis and strategic planning which would not have been possible otherwise. We present two approaches that drive a simulation to determine the quality of the generated solutions: an event-based approach and a fuzzy rule-based approach. While both of them are able to generate valid plans, the rule-based approach substantially outperforms the event-based one with respect to convergence time and quality of the solution.Sven Schellenberg, Arvind Mohais, Maksud Ibrahimov, Neal Wagner and Zbigniew Michalewic
Evolving greenfield passive optical networks
No full textWe investigate applying an evolutionary algorithm (EA) to the design of a passive optical network (PON). We use three techniques to improve the performance. Firstly, to reduce the risk of sub-optimal convergence, we use a novel genetic encoding. Secondly, we combine the EA with a heuristic to guide the optimisation. Thirdly, we investigate various ways of sub-dividing the problem. We briefly present experiments to demonstrate how the EA performs. The results show the strengths and weaknesses of the various techniques we employ
Static experts and dynamic enemies in coevolutionary games
No full textCopyright © 2007 IEEEThe usage of memory in coevolutionary systems offers a wide range of research possibilities, especially when evolving computationally intelligent computer players for games. The research discussed here continues from previous work done to include memory with coevolution for the game of TEMPO. The strategy of inserting a simple human derived rule base to kick start the evolutionary process with memory is investigated further, with tests done on the effectiveness of the expert as a participant in the evolutionary process. There is also further research presented on reproducing the human long term memory mechanism in the coevolutionary process, with a process used to mimic the way humans recall information relevant to the current scenario. This creates a memory that changes as the environmental situation changes, and results in a dynamic opposition to coevolve against.Avery, P.M.: Michalewicz, Z
The performance of an adaptive portfolio management system
No full textThis paper describes the operation and performance of a computational intelligence rule-base system that manages a portfolio of stocks according to investment objectives. We present an overview of several improvements to the system presented in previous papers and provide detailed results from applying the system in representative scenarios toward determining the robustness of the approach.Ghandar, A.; Michalewicz, Z.; Thuy-Duong To and Zurbruegg, R
Evolutionary Optimization
No full textThe emergence of different metaheuristics and their new variants in recent years has made the definition of the term Evolutionary Algorithms unclear. Originally, it was coined to put a group of stochastic search algorithms that mimic natural evolution together. While some people would still see it as a specific term devoted to this group of algorithms, including Genetic Algorithms, Genetic Programming, Evolution Strategies, Evolutionary Programming, and to a lesser extent Differential Evolution and Estimation of Distribution Algorithms, many others would regard “Evolutionary Algorithms” as a general term describing population-based search methods that involve some form of randomness and selection. In this chapter, we re-visit the fundamental question of “what is an Evolutionary Algorithm?” not only from the traditional viewpoint but also the wider, more modern perspectives relating it to other areas of Evolutionary Computation. To do so, apart from discussing the main characteristics of this family of algorithms we also look at Memetic Algorithms and the Swarm Intelligence algorithms. From our discussion, we see that establishing semantic borders between these algorithm families is not always easy, nor necessarily useful. It is anticipated that they will further converge as the research from these areas cross-fertilizes each other.Christian Blum, Raymond Chiong, Maurice Clerc, Kenneth De Jong, Zbigniew Michalewicz, Ferrante Neri and Thomas Weis
Experiments in applying evolutionary algorithms to software verification
No full textComplex concurrent systems present a significant challenge for software verification. If those systems are safety-critical, the need for software verification becomes particularly pressing, given the serious consequences of unforeseen defects. Complex concurrent systems are characterised by extremely large state spaces. The use of testing techniques for verification means that very little of the state space is explored. On the other hand, model-checking techniques exhaustively examine the state space, but will be stymied by the actual size. In this paper, we discuss some preliminary experiments on the application of evolutionary algorithms to software verification. This approach does not explore the whole state space, but does use heuristics to guide the search through the most promising parts of the state space for locating errors.Woei Shyang, Lakos, C., Michalewicz, Z. and Schellenberg, S
- …
