Search

The objective of this paper is to study the historical development of computer programmers for playing the game of checkers. Since the game-playing is a NP-hard problem, it would be interesting to use evolutionary algorithms to solve them. The question is can a programme be developed which can beat humans with complete success, it may appears that some challenges may also be formed which may substantiate the argument of the paper. Further, these challenges also form a part of this study. [ABSTRACT FROM AUTHOR]

Recently, Evolutionary Algorithms (EAs) with associative memory schemes have been developed to solve Dynamic Optimization Problems (DOPs). Current associative memory schemes always retrieve both the best memory individual and the corresponding environmental information. However, the memory individual with the best fitness could not be the most appropriate one for new environments. In this paper, two novel associative memory retrieving strategies are proposed to obtain the most appropriate memory environmental information. In these strategies, two best individuals are first selected from the two best memory individuals and the current best individual. Then, their corresponding environmental information is evaluated according to either the survivability or the diversity, one of which is retrieved. In experiments, the proposed two strategies were embedded into the state-of-the-art algorithm, i.e. the MPBIL, and tested on three dynamic functions in cyclic environments. Experiment results demonstrate that the proposed retrieving strategies enhance the search ability in cyclic environments. [ABSTRACT FROM AUTHOR]

This paper presents an application of Colonial Competitive Algorithm (CCA) in game theory and multi-objective optimization problems. The recently introduced CCA has proven its excellent capabilities, such as faster convergence and better global optimum achievement. In this paper CCA is used to find Nash Equilibrium points of nonlinear non-cooperative games. The proposed method can also be used as an alternative approach to solve multi-objective optimization problems. The effectiveness of the proposed method, in comparison to Genetic Algorithm, is proven through several static and dynamic example games and also multi-objective problems. [ABSTRACT FROM AUTHOR]

PRAM (Post Randomization Method) was introduced in 1997 but it is still one of the least used methods in statistical categorical data protection. This fact is because of the difficulty to obtain a good transition matrix in order to obtain a good protection. In this paper, we describe how to obtain a better protection using an evolutionary algorithm with integrated information loss and disclosure risk measures to find the best matrix. We also provide experiments using a real dataset of 1000 records in order to empirically evaluate the application of this technique. [ABSTRACT FROM AUTHOR]

In this paper, we evaluate parameter control strategies for evolutionary approaches to solve constrained combinatorial problems. For testing, we have used two well known evolutionary algorithms that solve the Constraint Satisfaction Problems GSA and SAW. We contrast our results with REVAC, a recently proposed technique for parameter tuning. [ABSTRACT FROM AUTHOR]

This paper presents the Gene Fragment Competition concept that can be used with Hybrid Genetic Algorithms specially in signal and image processing. Memetic Algorithms have shown great success in real-life problems by adding local search operators to improve the quality of the already achieved ˵good″ solutions during the evolutionary process. Nevertheless these traditional local search operators don΄t perform well in highly demanding evaluation processes. This stresses the need for a new semi-local non-exhaustive method. Our proposed approach sits as a tradeoff between classical Genetic Algorithms and traditional Memetic Algorithms, performing a quasi-global/quasi-local search by means of gene fragment evaluation and selection. The applicability of this hybrid Genetic Algorithm to the signal processing problem of Polyphonic Music Transcription is shown. The results obtained show the feasibility of the approach. [ABSTRACT FROM AUTHOR]

A possible alternative to topology fine-tuning for Neural Network (NN) optimization is to use Echo State Networks (ESNs), recurrent NNs built upon a large reservoir of sparsely randomly connected neurons. The promises of ESNs have been fulfilled for supervised learning tasks, but unsupervised ones, e.g. control problems, require more flexible optimization methods – such as Evolutionary Algorithms. This paper proposes to apply CMA-ES, the state-of-the-art method in evolutionary continuous parameter optimization, to the evolutionary learning of ESN parameters. First, a standard supervised learning problem is used to validate the approach and compare it to the standard one. But the flexibility of Evolutionary optimization allows us to optimize not only the outgoing weights but also, or alternatively, other ESN parameters, sometimes leading to improved results. The classical double pole balancing control problem is then used to demonstrate the feasibility of evolutionary (i.e. reinforcement) learning of ESNs. We show that the evolutionary ESN obtain results that are comparable with those of the best topology-learning methods. [ABSTRACT FROM AUTHOR]

One of the most important issues that must be taken in mind to optimize the design and the generalization abilities of trained artificial neural networks (ANN) is the architecture of the net. In this paper Symbiotic_RBF is proposed, a method to do automatically the process to design models for classification using symbiosis. For it, there are two populations who evolve together by means of coevolution. One of the populations is the method EvRBF, which provides the design of radial basis function neural nets by means of evolutionary algorithms. The second population evolves sets of parameters for the method EvRBF, being every individual of the population a configuration of parameters for the method. Thus, the main goal of Symbiotic_RBF is to find a suitable configuration of parameters necessary for the method EvRBF, which is adapted automatically to every problem. [ABSTRACT FROM AUTHOR]

The widespread use of artificial neural networks and the difficult work regarding the correct specification (tuning) of parameters for a given problem are the main aspects that motivated the approach purposed in this paper. This approach employs an evolutionary search to perform the simultaneous tuning of initial weights, transfer functions, architectures and learning rules (learning algorithm parameters). Experiments were performed and the results demonstrate that the method is able to find efficient networks with satisfactory generalization in a shorter search time. [ABSTRACT FROM AUTHOR]

This paper deals with the topic of learning through neuroevolutionary algorithms in non-stationary settings. This kind of algorithms that evolve the parameters and/or the topology of a population of Artificial Neural Networks have provided successful results in optimization problems in stationary settings. Their application to non-stationary problems, that is, problems that involve changes in the objective function, still requires more research. In this paper we address the problem through the integration of implicit, internal or genotypic, memory structures and external explicit memories in an algorithm called Promoter Based Genetic Algorithm with External Memory (PBGA-EM). The capabilities introduced in a simple genetic algorithm by these two elements are shown on different tests where the objective function of a problem is changed in an unpredictable manner. [ABSTRACT FROM AUTHOR]

Controlling a biped robot with a high degree of freedom to achieve stable movement patterns is still an open and complex problem, in particular within the RoboCup community. Thus, the development of control mechanisms for biped locomotion have become an important field of research. In this paper we introduce a model-free approach of biped motion generation, which specifies target angles for all driven joints and is based on a neural oscillator. It is potentially capable to control any servo motor driven biped robot, in particular those with a high degree of freedom, and requires only the identification of the robot΄s physical constants in order to provide an adequate simulation. The approach was implemented and successfully tested within a physical simulation of our target system - the 19-DoF Bioloid robot. The crucial task of identifying and optimizing appropriate parameter sets for this method was tackled using evolutionary algorithms. We could show, that the presented approach is applicable in generating walking patterns for the simulated biped robot. The work demonstrates, how the important parameters may be identified and optimized when applying evolutionary algorithms. Several so evolved controllers were capable of generating a robust biped walking behavior with relatively high walking speeds, even without using sensory information. In addition we present first results of laboratory experiments, where some of the evolved motions were tried to transfer to real hardware. [ABSTRACT FROM AUTHOR]

The art of solving the Mastermind puzzle was initiated by Donald Knuth and is already more than thirty years old; despite that, it still receives much attention in operational research and computer games journals, not to mention the natureinspired stochastic algorithm literature. In this paper we try to suggest a strategy that will allow nature-inspired algorithms to obtain results as good as those based on exhaustive search strategies; in order to do that, we first review, compare and improve current approaches to solving the puzzle; then we test one of these strategies with an estimation of distribution algorithm. Finally, we try to find a strategy that falls short of being exhaustive, and is then amenable for inclusion in nature inspired algorithms (such as evolutionary of particle swarm algorithms). This paper proves that by the incorporation of what we call local entropy into the fitness function of the evolutionary algorithm it becomes a better player than a random one, and gives a rule of thumb on how to incorporate the best heuristic strategies to evolutionary algorithms without incurring in an excessive computational cost. [ABSTRACT FROM AUTHOR]

This paper extends the current development of a methodology for Computer Aided Innovation. It begins with a presentation of concepts related to the perceived capabilities of virtual environments in the Innovation Cycle. The main premise establishes that it is possible to predict the acceptance of a new product in a specific market, by releasing an early prototype in a virtual scenario to quantify its general reception and to receive early feedback from potential customers. The paper continues to focus this research on a synergistic extension of techniques that have their origins in optimization and innovation disciplines. TRIZ (Theory of Inventive Problem Solving), extends the generation of variants with Evolutionary Algorithms (EA) and finally to present the designer and the intended customer, creative and innovative alternatives. All of this developed on a virtual software interface (Virtual World). The work continues with a general description of the project as a step forward to improve the overall strategy. [ABSTRACT FROM AUTHOR]

Even though in some instances optimisation can be computationally expensive, aeronautical industries are now considering with high interest the important cost reduction of introducing optimisation early in the design process. It has also been shown, in various UAVs and UCAVs applications that a multidisciplinary approach can provide additional benefits such as reduction of empty weight, drag and/or radar cross section. One of the new challenges in aeronautics is combining and accounting for multiple disciplines while considering uncertainty or variability in the design parameters or operating conditions. This paper describes a methodology for multidisciplinary design optimisation when there is uncertainty in the operating conditions. The methodology is based on canonical evolution algorithms and incorporates the concepts of multi-objective optimisation, hierarchical (multi-fidelity) topology, asynchronous evaluation and parallel computing (HAPMOEA). This methodology is enhanced by its coupling with an uncertainty analysis technique. The paper illustrates the use of this methodology on three practical test cases with increasing levels of complexity. Stealth aircraft and unmanned aerial vehicles are the ideal candidates due to the multi-physics involved and variability of Sectors to be performed. The first case considers the aerodynamic analysis and optimisation on a UCAV only, the second test compares and illustrates the challenge and benefits on introducing a second discipline (Electro-magnetic) while accounting for uncertainty in the designparameters and operating conditions. Results obtained from the optimisation show that the method is effective to find useful Pareto non-dominated solutions and the future benefit of using Uncertainty design technique. [ABSTRACT FROM AUTHOR]

In this paper a new hybrid approach that integrates an evolutionary algorithm with local search for microarray biclustering is presented. The novelty of this proposal is constituted by the incorporation of two mechanisms: the first one avoids loss of good solutions through generations and overcomes the high degree of overlap in the final population; and the other one preserves an adequate level of genotypic diversity. The performance of the memetic strategy was compared with the results of several salient biclustering algorithms over synthetic data with different overlap degrees and noise levels. In this regard, our proposal achieves results that outperform the ones obtained by the referential methods. Finally, a study on real data was performed in order to demonstrate the biological relevance of the results of our approach. [ABSTRACT FROM AUTHOR]

In this paper we present a control architecture for multi-robot systems in dynamic environments, where the low level behaviors are obtained through artificial neural networks and evolutionary algorithms to achieve collaborative behaviors in a multi-robot system. As an example, we have cooperative tasks establishing a surveillance scenario stressing cooperation and competition between them. [ABSTRACT FROM AUTHOR]

Recently, a new iterative optimization framework utilizing an evolutionary algorithm called ″Prototype Optimization with Evolved iMprovement Steps″ (POEMS) was introduced, which showed good performance on hard optimization problems - large instances of TSP and real-valued optimization problems. Especially, on discrete optimization problems such as the TSP the algorithm exhibited much better search capabilities than the standard evolutionary approaches. In many real-world optimization problems a solution is sought for multiple (conflicting) optimization criteria. This paper proposes a multiobjective version of the POEMS algorithm (mPOEMS), which was experimentally evaluated on the multiobjective 0/1 knapsack problem with alternative multiobjective evolutionary algorithms. Major result of the experiments was that the proposed algorithm performed comparable to or better than the alternative algorithms. [ABSTRACT FROM AUTHOR]

In the current paper we present an integrated genetic programming environment with a graphical user interface (GUI), called jGPModeling. The jGPModeling environment was developed using the JAVA programming language, and is an implementation of the steady-state genetic programming algorithm. That algorithm evolves tree based structures that represent models of input – output relation of a system. During the design and implementation of the application, we focused on the execution time optimization and tried to limit the bloat effect. In order to evaluate the performance of the jGPModeling environment, two different real world system modeling tasks were used. [ABSTRACT FROM AUTHOR]

The paper explores connections between population topology and individual interactions inducing autonomy, communication and learning. A Collaborative Asynchronous Multi-Population Evolutionary (CAME) model is proposed. Each individual in the population acts as an autonomous agent with the goal of optimizing its fitness being able to communicate and select a mate for recombination. Different strategies for recombination correspond to different societies of agents (subpopulations). The asynchronous search process is facilitated by a gradual propagation of the fittest individuals΄ genetic material into the population. Furthermore, two heuristics are proposed for avoiding local optima and for maintaining population diversity. These are the dynamic dominance heuristic and the shaking mechanism, both being integrated in the CAME model. Numerical results indicate a good performance of the proposed evolutionary asynchronous search model. Particularly, proposed CAME technique obtains excellent results for difficult highly multimodal optimization problems indicating a huge potential for dynamic and multicriteria optimization. [ABSTRACT FROM AUTHOR]

We present a multi-objective approach to tackle a real-world nurse scheduling problem using an evolutionary algorithm. The aim is to generate a few good quality non-dominated schedules so that the decision-maker can select the most appropriate one. Our approach is designed around the premise of `satisfying individual nurse preferences΄ which is of practical significance in our problem. We use four objectives to measure the quality of schedules in a way that is meaningful to the decision-maker. One objective represents staff satisfaction and is set as a target. The other three objectives, which are subject to optimisation, represent work regulations and workforce demand. Our algorithm incorporates a self-adaptive decoder to handle hard constraints and a re-generation strategy to encourage production of new genetic material. Our results show that our multi-objective approach produces good quality schedules that satisfy most of the nurses΄ preferences and comply with work regulations and workforce demand. The contribution of this paper is in presenting a multi-objective evolutionary algorithm to nurse scheduling in which increasing overall nurses΄ satisfaction is built into the self-adaptive solution method. [ABSTRACT FROM AUTHOR]

In recent years we have witnessed a tremendous growth of communication networks resulted in a large variety of combinatorial optimization problems. One of these problems is the terminal assignment problem. In this paper, we propose a genetic algorithm employing multiple crossover and mutation operators for solving the well-known terminal assignment problem. Two sets of available crossover and mutation operators are established initially. In each generation a crossover method is selected for recombination and a mutation method is selected for mutation based on the amount fitness improvements achieved over a number of previous operations (recombinations/mutations). We use tournament selection for this purpose. Simulation results with the different methods implemented are compared. [ABSTRACT FROM AUTHOR]

This paper presents a brief history of the western music: from its genesis to serialism and the Darmstadt school. Also some mathematical aspects of music are then presented and confronted with music as a form of art. The question is, are these two distinct aspects compatible? Can computers be of real help in automatic composition? The more appealing algorithmic approach is evolutionary computation as it offers creativity potential. Therefore, the Evolutionary Algorithms are then introduced and some results of GAs and GPs application to music generation are analysed. [ABSTRACT FROM AUTHOR]

Evolutionary algorithms have received much attention in extracting knowledge on high-dimensional micro-array data, being crucial to their success a suitable definition of the search space of the potential solutions. In this paper, we present an evolutionary approach for selecting informative genes (features) to predict and diagnose cancer. We propose a procedure that combines results of filter methods, which are commonly used in the field of data mining, to reduce the search space where a genetic algorithm looks for solutions (i.e. gene subsets) with better classification performance, being the quality (fitness) of each solution evaluated by a classification method. The methodology is quite general because any classification algorithm could be incorporated as well a variety of filter methods. Extensive experiments on a public micro-array dataset are presented using four popular filter methods and SVM. [ABSTRACT FROM AUTHOR]

This paper proposes the use of a parallelmultiobjective evolutionary technique to predict patterns, motifs, in real deoxyribonucleic acid (DNA) sequences. DNA analysis is a very important branch within bioinformatics, resulting in a large number of NP-hard optimization problems such as multiple alignment, motif finding, or protein folding. In this work we study the use of amultiobjective evolutionary algorithms team to solve the Motif Discovery Problem. According to this, we have designed a parallel heuristic that allows the collaborative work of four algorithms, two population-based algorithms: Differential Evolution with Pareto Tournaments and Nondominated Sorting Genetic Algorithm II, and two trajectory-based algorithms: Multiobjective Variable Neighborhood Search and Multiobjective Skewed Variable Neighborhood Search. In this way, we take advantage of the properties of different algorithms, getting to expand the search space covered in our problem. As we will see, the results obtained by our team significantly improve the results published in previous research. [ABSTRACT FROM AUTHOR]

Real-world classification problems usually involve imbalanced data sets. In such cases, a classifier with high classification accuracy does not necessarily imply a good classification performance for all classes. The Area Under the ROC Curve (AUC) has been recognized as a more appropriate performance indicator in such cases. Quite a few methods have been developed to design classifiers with the maximum AUC. In the context of Neural Networks (NNs), however, it is usually an approximation of AUC rather than the exact AUC itself that is maximized, because AUC is non-differentiable and cannot be directly maximized by gradient-based methods. In this paper, we propose to use evolutionary algorithms to train NNs with the maximum AUC. The proposed method employs AUC as the objective function. An evolutionary algorithm, namely the Self-adaptive Differential Evolution with Neighborhood Search (SaNSDE) algorithm, is used to optimize the weights of NNs with respect to AUC. Empirical studies on 19 binary and multi-class imbalanced data sets show that the proposed evolutionary AUC maximization (EAM) method can train NN with larger AUC than existing methods. [ABSTRACT FROM AUTHOR]

This paper analyzes the non-linear dynamics of a MEMS Gyroscope system, modeled with a proof mass constrained to move in a plane with two resonant modes, which are nominally orthogonal. The two modes are ideally coupled only by the rotation of the gyro about the plane΄s normal vector. We demonstrated that this model has an unstable behavior. Control problems consist of attempts to stabilize a system to an equilibrium point, a periodic orbit, or more general, about a given reference trajectory. We also developed a particle swarm optimization technique for reducing the oscillatory movement of the nonlinear system to a periodic orbit. [ABSTRACT FROM AUTHOR]

In the current paper is presented the application of a Gene Expression Programming Environment in modeling the fatigue behavior of composite materials. The environment was developed using the JAVA programming language, and is an implementation of a variation of Gene Expression Programming. Gene Expression Programming (GEP) is a new evolutionary algorithm that evolves computer programs (they can take many forms: mathematical expressions, neural networks, decision trees, polynomial constructs, logical expressions, and so on). The computer programs of GEP, irrespective of their complexity, are all encoded in linear chromosomes. Then the linear chromosomes are expressed or translated into expression trees (branched structures). Thus, in GEP, the genotype (the linear chromosomes) and the phenotype (the expression trees) are different entities (both structurally and functionally). This is the main difference between GEP and classical tree based Genetic Programming techniques. In order to evaluate the performance of the presented environment, we tested it in fatigue modeling of composite materials. [ABSTRACT FROM AUTHOR]

This paper introduces a hyper-heuristic approach for the Unit Commitment Problem (UCP). Tests are performed using benchmark data from literature and real-world data from the Turkish interconnected power network. The proposed hyper-heuristic and several methods applied previously to the UCP, are compared. Results show that the hyper-heuristic method achieves good results in all test sets. Furthermore, it is also a robust method for increased problem sizes without the need for parameter tuning. Based on the promising results, research will continue for further improvements. [ABSTRACT FROM AUTHOR]

In this paper we present an evolutionary algorithm to tackle the aggregation network design in a mobile communication system. The design and optimization of this part of the network involves four different tasks: the determination of the number and location of the Base Station Controller (BSC) or Radio Network Controllers (RNC), the assignment of Base Stations (BTS) or B-Nodes to the controllers, the definition of the tree structure that links all the nodes with the controllers and, finally, the system assignment in the links between the different hops of the tree. The novel evolutionary heuristic proposed deals with all these sub-problem together and it is able to obtain good solutions, as will be shown in several real scenarios. [ABSTRACT FROM AUTHOR]

Evolutionary algorithms can be used to evolve novel digital circuit solutions. This paper proposes the use of flexible target truth tables, allowing evolution more freedom where values are undefined. This concept is applied to three test circuits with different distributions of ˵don΄t care″ values. Two strategies are introduced for utilising the undefined output values within the evolutionary algorithm. The use of flexible desired truth tables is shown to significantly improve the success of the algorithm in evolving circuits to perform this function. In addition, we show that this flexibility allows evolution to develop more hardware efficient solutions than using a fully-defined truth table. [ABSTRACT FROM AUTHOR]

This paper describes experience with NEAT (NeuroEvolution of Augmenting Topologies) method which is based on evolutionary computation and optimization of neural networks structure and synaptic weights. Non-linear function XOR approximation is tested and evaluated with this method with the aim of perspective application in humanoid robot NAO. The experiments show that selected method NEAT is suitable for this type of adaptation of NN, because of its ability to deal with the problems which emerge in TWEAN methods. [ABSTRACT FROM AUTHOR]

This paper studies how evolutionary algorithms (EA) scale with growing genome size, when used for similarity-based clustering. A simple EA and EAs with problem-dependent knowledge are experimentally evaluated for clustering up to 100,000 objects. We find that EAs with problem-dependent crossover or hybridization scale near-linear in the size of the similarity matrix, while the simple EA, even with problem-dependent initialization, fails at moderately large genome sizes. [ABSTRACT FROM AUTHOR]

Direct marketing modeling identifies effective models for improving managerial decision making in marketing. This paper proposes a novel system for discovering models represented as Bayesian networks from incomplete databases in the presence of missing values. It combines an evolutionary algorithm with the traditional Expectation-Maximization(EM) algorithm to find better network structures in each iteration round. A data completing method is also presented for the convenience of learning and evaluating the candidate networks. The new system can overcome the problem of getting stuck in sub-optimal solutions which occurs in most existing learning algorithms and the efficiency problem in some existing evolutionary algorithms. We apply it to a real-world direct marketing modeling problem, and compare the performance of the discovered Bayesian networks with other models obtained by other methods. In the comparison, the Bayesian networks learned by our system outperform other models. [ABSTRACT FROM AUTHOR]

This paper introduces DAFO, a Distributed Agent Framework for Optimization that helps in designing and applying Coevolutionary Genetic Algorithms (CGAs). CGAs have already proven to be efficient in solving hard optimization problems, however they have not been considered in the existing agent-based metaheuristics frameworks that currently provide limited organization models. As a solution, DAFO includes a complete organization and reorganization model, Multi-Agent System for EVolutionary Optimization (MAS4EVO), that permits to formalize CGAs structure, interactions and adaptation. Examples of existing and original CGAs modeled using MAS4EVO are provided and an experimental proof of their efficiency is given on an emergent topology control problem in mobile hybrid ad hoc networks called the injection network problem. [ABSTRACT FROM AUTHOR]

Among the penalty based approaches for constrained optimization, Augmented Lagrangian (AL) methods are better in at least three ways: (i) they have theoretical convergence properties, (ii) they distort the original objective function minimally to allow a better search behavior, and (iii) they can find the optimal Lagrange multiplier for each constraint as a by-product of optimization. Instead of keeping a constant penalty parameter throughout the optimization process, these algorithms update the parameters adaptively so that the corresponding penalized function dynamically changes its optimum from the unconstrained minimum point to the constrained minimum point with iterations. However, the flip side of these algorithms is that the overall algorithm is a serial implementation of a number of optimization tasks, a process that is usually time-consuming. In this paper, we devise a genetic algorithm based parameter update strategy to a particular AL method. The strategy is self-adaptive in order to make the overall genetic algorithm based augmented Lagrangian (GAAL) method parameter-free. The GAAL method is applied to a number of constrained test problems taken from the EA literature. The function evaluations required by GAAL in many problems is an order or more lower than existing methods. [ABSTRACT FROM AUTHOR]

In a bi-criteria optimization problem, often the user is interested in a subset of solutions lying in the knee region. On the other hand in many problem-solving tasks, often one or a few methodologies are commonly used. In this paper, we argue that there is a link between the knee solutions in bi-criteria problems and the preferred methodologies when viewed from a conflicting bi-criterion standpoint. We illustrate our argument with the help of a number of popularly used problem-solving tasks. Each task, when perceived as a bicriteria problem, seems to exhibit a knee or a knee-region and the commonly-used methodology seems to lie within the knee-region. This linking is certainly an interesting finding and may have a long-term implication in the development of efficient solution methodologies for different scientific and other problem-solving tasks. [ABSTRACT FROM AUTHOR]

There are numerous algorithms available for training artificial neural networks. Besides classical algorithms for supervised learning such as backpropagation, associative memory and radial basis function, this training task can be employed by evolutionary computation since most of the gradient descent related algorithms can be view as an application of optimization theory and stochastic search. In this paper, the logistic model of population growth from ecology is integrated into initialization, selection and crossover operators of genetic algorithms for neural network training. These chaotic operators are very efficient in maintaining the population diversity during the evolution process of genetic algorithms. A comparison is done on the basis of a benchmark comprising several data classification problems for neural networks. Three variants of training – Backpropagation (BP), Genetic Algorithms (GA) and Genetic Algorithms with Chaotic Operators (GACO) – are described and compared. The experimental results confirm the dynamic mobility of chaotic algorithms in GACO network training, which can overcome saturation and improve the convergence rate. [ABSTRACT FROM AUTHOR]

In this paper, a new memetic approach that integrates a Multi-Objective Evolutionary Algorithm (MOEA) with local search for microarray biclustering is presented. The original features of this proposal are the consideration of opposite regulation and incorporation of a mechanism for tuning the balance between the size and row variance of the biclusters. The approach was developed according to the Platform and Programming Language Independent Interface for Search Algorithms (PISA) framework, thus achieving the possibility of testing and comparing several different memetic MOEAs. The performance of the MOEA strategy based on the SPEA2 performed better, and its resulting biclusters were compared with those obtained by a multi-objective approach recently published. The benchmarks were two datasets corresponding to Saccharomyces cerevisiae and human B-cells Lymphoma. Our proposal achieves a better proportion of coverage of the gene expression data matrix, and it also obtains biclusters with new features that the former existing evolutionary strategies can not detect. [ABSTRACT FROM AUTHOR]

This paper presents a novel representation of Cartesian genetic programming (CGP) in which multiple networks are used in the classification of high resolution X-rays of the breast, known as mammograms. CGP networks are used in a number of different recombination strategies and results are presented for mammograms taken from the Lawrence Livermore National Laboratory database. [ABSTRACT FROM AUTHOR]

The dynamic, multi-dimensional knapsack problem is an important benchmark for evaluating the performance of evolutionary algorithms in changing environments, especially because it has many real-world applications. In order to analyze the performance of an evolutionary algorithm according to this benchmark, one needs to be able to change the current problem in a controlled manner. Several methods have been proposed to achieve this goal. In this paper, we briefly outline the proposed methods, discuss their shortcomings and propose a new method that can generate changes for a given severity level more reliably. We then present the experimental setup and results for the new method and compare it with existing methods. The current results are promising and promote further study. [ABSTRACT FROM AUTHOR]

A Supply Chain (SC) is a complex network of facilities with dissimilar and conflicting objectives, immersed in an unpredictable environment. Discrete-event simulation is often used to model and capture the dynamic interactions occurring in the SC and provide SC performance indicators. However, a simulator by itself is not an optimizer. This paper therefore considers the hybridization of Evolutionary Algorithms (EAs), well known for their multi-objective capability, with an SC simulation module in order to determine the inventory policy (order-point or order-level) of a single product SC, taking into account two conflicting objectives: the maximization of customer service level and the total inventory cost. Different evolutionary approaches, such as SPEA-II, SPEA-IIb, NSGA-II and MO-PSO, are tested in order to decide which algorithm is the most suited for simulation-based optimization. The research concludes that SPEA-II favors a rapid convergence and that variation and crossover schemes play and important role in reaching the true Pareto front in a reasonable amount of time. [ABSTRACT FROM AUTHOR]

In this paper, a symbolic algorithm for solving constrained multi-objective optimisation problems is proposed. It is used to get the Pareto optimal solutions as functions of KKT multipliers ]> for multi-objective problems with continuous, differentiable, and convex/pseudo-convex functions. The algorithm is able to detect the relationship between the decision variables that form the exact curve/hyper-surface of the Pareto front. This algorithm enables to formulate an analytical form for the true Pareto front which is necessary in absolute performance measurement of evolutionary computing techniques. Here the proposed technique is tested on some test problems which have been chosen from a number of significant past studies. The results show that the proposed symbolic algorithm is robust to find the analytical formula of the exact Pareto front. [ABSTRACT FROM AUTHOR]

Traditionally, Multiobjective Evolutionary Algorithms (MOEAs) aim at approximating the entire true pareto-front of their input problems. However, the actual number of solutions with different trade-offs between objectives in a resulting pareto-front is often too large to be applicable in practice. The new field Multiobjective Distinct Candidates Optimization (MODCO) research is concerned with the optimization of a low and user-defined number of clearly distinct candidates. This dramatically decreases the amount of post-processing needed in the decision making process of which solution to actually implement, as described in our related technical repport ˵Multiobjective Distinct Candidates Optimization (MODCO): A new Branch of Multiobjective Optimization Research″ [9]. In this paper, we introduce the first algorithm designed for the challenges of MODCO; providing a given number of distinct solutions as close as possible to the true pareto-front. The algorithm is using subpopulations to enforce clusters of solutions, in such a way that the number of clusters formed can be set directly. The algorithm is based on the Differential Evolution for Multiobjective Optimization (DEMO) algorithm versions, but is exchanging the crowding/density measure with two alternating secondary fitness measures. Applying these measures ensures that subpopulations are attracted towards knee regions while also making them repel each other if they get too close to one another. This way subpopulations traverse different parts of the objective space while forming clusters each returning a single distinct solution. [ABSTRACT FROM AUTHOR]

The main purpose of this study is to find out a good solution to the vehicle routing problem considering heterogeneous vehicles. This problem tries to solve the generation of paths and the assignment of buses on these routes. The objective of this problem is to minimize the number of vehicles required and to maximize the number of demands transported. This paper considers a Memetic Algorithm for the vehicle routing problem with heterogeneous fleet for any transport problem between many origins and many destinations. A Memetic Algorithm always maintains a population of different solutions to the problem, each of which operates as an agent. These agents interact between themselves within a framework of competition and cooperation. Extensive computational tests on some instances taken from the literature reveal the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

The design of intelligent controllers for nonlinear systems continues to be a challenging problem, particularly when the system is uncertain or the environment noisy. A nonparametric approach which has gained success is to employ a neural network to learn about the unknown plant and fuzzy inference to compensate for the uncertainty (GANFIS control). Inherent in the design of such controllers is the need to tune the weights of the GANFIS controller. Evolutionary learning has been suggested to tune the GANFIS parameters but a difficulty is selecting the parameters for tuning. Further, it is well known that proper selection of the fitness function has an important effect on system performance. In this paper, we integrate two design techniques that we have previously developed into a single generalized ANFIS controller: adaptive tuners to select critical evolutionary parameters and a predictive fitness function for measuring system performance. The adaptive tuners also employ this predictive fitness as part of selection process which is a new approach. Results show that this approach is a feasible method in designing GANFIS controllers using evolutionary tuning and predictive fitness. [ABSTRACT FROM AUTHOR]

A facial composite generated from an eyewitness΄s memory often constitutes the first and only means available for police forces to identify a criminal suspect. To date, commercial computerised systems for constructing facial composites have relied almost exclusively on a feature-based, `cut-andpaste΄ method whose effectiveness has been fundamentally limited by both the witness΄s limited ability to recall and verbalise facial features and by the large dimensionality of the search space. We outline a radically new approach to composite generation which combines a parametric, statistical model of facial appearance with a computational search algorithm based on interactive, evolutionary principles. We describe the fundamental principles on which the new system has been constructed, outline recent innovations in the computational search procedure and also report on the real-world experience of UK police forces who have been using a commercial version of the system. [ABSTRACT FROM AUTHOR]

Logistic regression (LR) is a simple and efficient supervised learning algorithm for estimating the probability of an outcome variable. This algorithm is widely accepted and used in medicine for classification of diseases using DNA microarray data. Classical LR does not perform well for microarrays when applied directly, because the number of variables exceeds the number of samples. However, by reducing the number of genes and selecting specific variables (using filtering methods) great results can be obtained with this algorithm. On this contribution we propose a novel approach for fitting the (penalized) LR models based on EDAs. Breast Cancer dataset has been proposed to compare both accuracy and gene selection. [ABSTRACT FROM AUTHOR]

This work introduces SymbPar, a parallel meta-evolutionary algorithm designed to build Radial Basis Function Networks minimizing the number of parameters needed to be set by hand. Parallelization is implemented using independent agents to evaluate every individual. Experiments over classifications problems show that the new method drastically reduces the time took by sequential algorithms, while maintaining the generalization capabilities and sizes of the nets it builds. [ABSTRACT FROM AUTHOR]

The Differential Evolution is a floating-point evolutionary algorithm that has demonstrated good performance on locating the global optima in a wide variety of problems and applications. It has mainly three tuning parameters and their choice is fundamental to ensure good quality solutions. Because of this, adaptive parameter control and self-adaptive parameter control had been object of research. We present a novel scheme for controlling two parameters of the Differential Evolution using fitness information of the population in each generation. The algorithm shows outstanding performance on a well known benchmark functions, improving the standard DE and comparable with similar algorithms. [ABSTRACT FROM AUTHOR]

In this work we examine the effect of elitist and non-elitist selection on a supply chain problem. The problem is characterized by an output constraint which in turn separates the search space in a feasible and a non-feasible region. Additionally the simulation output is noisy due to a stochastic demand model. We will show analyze which strategy is able to perform a walk on the boundary between the feasible and infeasible space. Additionally a new selection scheme is introduced based on a statistical test to evaluate the difference between two solutions given a number of noisy quality values. This selection scheme is described and evaluated on the problem situation. [ABSTRACT FROM AUTHOR]