Search

This paper describes a class of random additively decomposable problems (rADPs) with and without interactions between the subproblems. The paper then tests the hierarchical Bayesian optimization algorithm (hBOA) and other evolutionary algorithms on a large number of random instances of the proposed class of problems. The results show that hBOA can scalably solve rADPs and that it significantly outperforms all other methods included in the comparison. Furthermore, the results provide a number of interesting insights into both the difficulty of a broad class of decomposable problems as well as the sensitivity of various evolutionary algorithms to different sources of problem difficulty. rADPs can be used to test other optimization algorithms. [ABSTRACT FROM AUTHOR]

The bin-packing problem is a well known NP-Hard optimisation problem, and, over the years, many heuristics have been developed to generate good quality solutions. This paper outlines a genetic programming system which evolves a heuristic that decides whether to put a piece in a bin when presented with the sum of the pieces already in the bin and the size of the piece that is about to be packed. This heuristic operates in a fixed framework that iterates through the open bins, applying the heuristic to each one, before deciding which bin to use. The best evolved programs emulate the functionality of the human designed ‘first-fit' heuristic. Thus, the contribution of this paper is to demonstrate that genetic programming can be employed to automatically evolve bin packing heuristics which are the same as high quality heuristics which have been designed by humans. [ABSTRACT FROM AUTHOR]

Under mild conditions, the Pareto set of a continuous multi-objective optimization problem exhibits certain regularity. We have recently advocated taking into consideration such regularity in designing multi-objective evolutionary algorithms. Following our previous work on using Local Principal Component Analysis for capturing the regularity, this paper presents a new approach for acquiring and using the regularity of the Pareto set in evolutionary algorithms. The approach is based on the Generative Topographic Mapping and can be regarded as an Estimation of Distribution Algorithm. It builds models of the distribution of promising solutions based on regularity patterns extracted from the previous search, and samples new solutions from the models thus built. The proposed algorithm has been compared with two other state-of-the-art algorithms, NSGA-II and SPEA2 on a set of test problems. [ABSTRACT FROM AUTHOR]

In this paper we describe a computationally intelligent approach to solving the dynamic vehicle routing problem where a fleet of vehicles needs to be routed to pick up loads at customers and drop them off at a depot. Loads are introduced online during the actual planning of the routes. The approach described in this paper uses an evolutionary algorithm (EA) as the basis of dynamic optimization. For enhanced performance, not only are currently known loads taken into consideration, also possible future loads are considered. To this end, a probabilistic model is built that describes the behavior of the load announcements. This allows the routing to make informed anticipated moves to customers where loads are expected to arrive shortly. Our approach outperforms not only an EA that only considers currently available loads, it also outperforms a recently proposed enhanced EA that performs anticipated moves but doesn't employ explicit learning. Our final conclusion is that under the assumption that the load distribution over time shows sufficient regularity, this regularity can be learned and exploited explicitly to arrive at a substantial improvement in the final routing efficiency. [ABSTRACT FROM AUTHOR]

This paper analyzes the hierarchical Bayesian optimization algorithm (hBOA) on the problem of finding ground states of Ising spin glasses with ±J couplings in two and three dimensions. The performance of hBOA is compared to that of the simple genetic algorithm (GA) and the univariate marginal distribution algorithm (UMDA). The performance of all tested algorithms is improved by incorporating a deterministic hill climber (DHC) based on single-bit flips and cluster exact approximation (CEA). The results show that hBOA significantly outperforms GA and UMDA with both types of local search and that CEA enables all tested algorithms to solve larger spin-glass instances than DHC. Using advanced hybrid methods created by combining competent genetic and evolutionary algorithms with advanced local searchers thus proves advantageous in this challenging class of problems. [ABSTRACT FROM AUTHOR]

Geometric crossover is a representation-independent definition of crossover based on the distance of the search space interpreted as a metric space. It generalizes the traditional crossover for binary strings and other important recombination operators for the most frequently used representations. Using a distance tailored to the problem at hand, the abstract definition of crossover can be used to design new problem specific crossovers that embed problem knowledge in the search. In this paper, we introduce the important notion of product geometric crossover that allows to build new geometric crossovers combining pre-existing geometric crossovers in a simple way. [ABSTRACT FROM AUTHOR]

Support Vector Machines (SVMs) are well-established Machine Learning (ML) algorithms. They rely on the fact that i) linear learning can be formalized as a well-posed optimization problem; ii) non-linear learning can be brought into linear learning thanks to the kernel trick and the mapping of the initial search space onto a high dimensional feature space. The kernel is designed by the ML expert and it governs the efficiency of the SVM approach. In this paper, a new approach for the automatic design of kernels by Genetic Programming, called the Evolutionary Kernel Machine (EKM), is presented. EKM combines a well-founded fitness function inspired from the margin criterion, and a co-evolution framework ensuring the computational scalability of the approach. Empirical validation on standard ML benchmark demonstrates that EKM is competitive using state-of-the-art SVMs with tuned hyper-parameters. [ABSTRACT FROM AUTHOR]

Understanding how neutrality works in EC systems has drawn increasing attention. However, some researchers have found neutrality to be beneficial for the evolutionary process while others have found it either useless or worse. We believe there are various reasons for these contradictory results: (a) many studies have based their conclusions using crossover and mutation as main operators rather than using only mutation (Kimura's studies were done analysing only mutations) and, (b) studies often consider problems and representation with larger complexity. The aim of this paper is to analyse how neutral mutations tend to behave in GP and establish how important they are. For this purpose we introduce an approach which has two advantages: (a) it allows us to specify neutrality and, (b) this makes possible to understand how neutrality affects the evolutionary search process. [ABSTRACT FROM AUTHOR]

The effects of neutrality on evolutionary search have been considered in a number of interesting studies, the results of which, however, have been contradictory. We believe that this confusion is due to several reasons. In this paper, we shed some light on neutrality by addressing these problems. That is, we use the simplest possible definition of neutrality, we consider one of the simplest possible algorithms, we apply it to two problems (a unimodal landscape and a deceptive landscape), which we analyse using fitness distance correlation, performance statistics and, critically, tracking the full evolutionary path of individuals within their family tree. [ABSTRACT FROM AUTHOR]

In this paper, we present a different look on splicing systems, namely as problem solvers. After defining the concept of accepting splicing system we discuss how these systems can be used as problem solvers. Then we construct an accepting splicing system able to uniformly solve SAT in time O(m+n) for a formula of length m over n variables. We also propose a uniform solution based on accepting splicing systems to HPP that runs in time O(n), where n is the number of vertices of the instance of HPP. [ABSTRACT FROM AUTHOR]

This paper presents studies on the life history evolution of plants carried out by experimenting with a multi-agent platform of generic virtual plants. The conducted simulations address the trade-off between resource allocation to vegetative and reproductive structures. The trade-off is pointed out by evolutionary runs selecting for one of the two traits. It is further shown that the introduction of an age at maturity is an effective measure to enhance both life history traits. A third series of experiments highlights that competition in plant communities has an impact on the trade-off. Depending on the competitive pressure, plants evolve more investment of resources into growth than into reproduction. The results corroborate some hypotheses of life history theory. [ABSTRACT FROM AUTHOR]

The major contribution of this paper is a novel security framework, which is inspired by the principles of Artificial Immune Systems (AIS), for Nature inspired routing protocols in general and for BeeHive in particular. We have designed an empirical validation framework to demonstrate that the new framework provides the same security level as BeeHiveGuard, a digital signature based cryptography framework. But the processing and communication costs of the new framework are significantly smaller as compared to BeeHiveGuard. [ABSTRACT FROM AUTHOR]

Swarm Intelligence is the emergent collective intelligence of groups of simple agents acting almost independently. Algorithms following this paradigm have many desirable properties: flexibility, decentralized control, robustness, and fault tolerance. This paper presents a novel agent coordination model inspired by the way ants collectively transport large preys. In our model a swarm of agents, each having a different destination to reach, moves with no centralized control in the direction indicated by the majority of agents keeping its initial shape. The model is used to build an algorithm for the problems of image alignment and image matching. The novelty of the approach and its effectiveness are discussed. [ABSTRACT FROM AUTHOR]

Linear Linkage Encoding (LLE) is a representational scheme proposed for Genetic Algorithms (GA). LLE is convenient to be used for grouping problems and it doesn't suffer from the redundancy problem that exists in classical encoding schemes. Any number of groups can be represented in a fixed length chromosome in this scheme. However, the length of the chromosome in LLE is determined by the number of elements to be grouped just like the other encoding schemes. This disadvantage becomes dominant when LLE is applied on large datasets and the encoding turns out to be an infeasible model. In this paper a two-level approach is proposed for LLE in order to overcome the problem. In this method, the large dataset is divided into a group of subsets. In the first phase of the process, the data in the subsets are grouped using LLE. Then these groups are used to obtain the final partitioning of the data in the second phase. The approach is tested on the clustering problem. Two considerably large datasets have been chosen for the experiments. It is not possible to obtain a satisfactory convergence with the straightforward application of LLE on these datasets. The method proposed can cluster the datasets with low error rates. [ABSTRACT FROM AUTHOR]

This paper describes and analyzes a series of experiments to develop a general evolutionary behavior acquisition technique for humanoid robots. The robot's behavior is defined by joint controllers evolved concurrently. Each joint controller consists of a series of primitive actions defined by a chromosome. By using genetic algorithms with specifically designed genetic operators and novel representations, complex behaviors are evolved from the primitive actions defined. Representations are specifically tailored to be useful in trajectory generation for humanoid robots. The effectiveness of the method is demonstrated by two experiments: a handstand and a limbo dance behavioral tasks (leaning the body backwards so as to pass under a fixed height bar). [ABSTRACT FROM AUTHOR]

Although a number of multiobjective evolutionary algorithms have been proposed over the last two decades, not much effort has been made to deal with variable linkages in multiobjective optimization. Recently, we have suggested a general framework of multiobjective evolutionary algorithms based on decomposition (MOEA/D) [1]. MOEA/D decomposes a MOP into a number of scalar optimization subproblems by a conventional decomposition method. The optimal solution to each of these problems is a Pareto optimal solution to the MOP under consideration. An appropriate decomposition could make these individual Pareto solutions evenly distribute along the Pareto optimal front. MOEA/D aims at solving these scalar optimization subproblems simultaneously. In this paper, we propose, under the framework of MOEA/D, a multiobjective differential evolution based decomposition (MODE/D) for tackling variable linkages. Our experimental results show that MODE/D outperforms several other MOEAs on several test problems with variable linkages. [ABSTRACT FROM AUTHOR]

In real-world optimisation problems, feasibility of solutions is invariably an essential requirement. A natural way to deal with feasibility is to cast it as an additional objective in a multi-objective optimisation setting. In this paper, we consider two possible ways to do this, using a multi-level scheme for ranking solutions. One strategy considers feasibility first, before considering objective values, while the other reverses this ordering. The first strategy has been explored before, while the second has not. Experiments show that the second strategy can be much more successful on some difficult problems. [ABSTRACT FROM AUTHOR]

In this paper, we propose the use of a mathematical programming technique called the ε-constraint method, hybridized with an evolutionary single-objective optimizer: the cultured differential evolution. The ε-constraint method uses the cultured differential evolution to produce one point of the Pareto front of a multiobjective optimization problem at each iteration. This approach is able to solve difficult multiobjective problems, relying on the efficiency of the single-objective optimizer, and on the fact that none of the two approaches (the mathematical programming technique or the evolutionary algorithm) are required to generate the entire Pareto front at once. The proposed approach is validated using several difficult multiobjective test problems, and our results are compared with respect to a multi-objective evolutionary algorithm representative of the state-of-the-art in the area: the NSGA-II. [ABSTRACT FROM AUTHOR]

This paper proposes an idea of probabilistically using a scalarizing fitness function in evolutionary multiobjective optimization (EMO) algorithms. We introduce two probabilities to specify how often the scalarizing fitness function is used for parent selection and generation update in EMO algorithms. Through computational experiments on multiobjective 0/1 knapsack problems with two, three and four objectives, we show that the probabilistic use of the scalarizing fitness function improves the performance of EMO algorithms. In a special case, our idea can be viewed as the probabilistic use of an EMO scheme in single-objective evolutionary algorithms (SOEAs). From this point of view, we examine the effectiveness of our idea. Experimental results show that our idea improves not only the performance of EMO algorithms for multiobjective problems but also that of SOEAs for single-objective problems. [ABSTRACT FROM AUTHOR]

This paper presents a new multi-objective evolutionary algorithm which consists of a hybrid between a particle swarm optimization approach and some concepts from rough sets theory. The main idea of the approach is to combine the high convergence rate of the particle swarm optimization algorithm with a local search approach based on rough sets that is able to spread the nondominated solutions found, so that a good distribution along the Pareto front is achieved. Our proposed approach is able to converge in several test functions of 10 to 30 decision variables with only 4,000 fitness function evaluations. This is a very low number of evaluations if compared with today's standards in the specialized literature. Our proposed approach was validated using nine standard test functions commonly adopted in the specialized literature. Our results were compared with respect to a multi-objective evolutionary algorithm that is representative of the state-of-the-art in the area: the NSGA-II. [ABSTRACT FROM AUTHOR]

Pattern classification seeks to minimize error of unknown patterns, however, in many real world applications, type I (false positive) and type II (false negative) errors have to be dealt with separately, which is a complex problem since an attempt to minimize one of them usually makes the other grow. Actually, a type of error can be more important than the other, and a trade-off that minimizes the most important error type must be reached. Despite the importance of type-II errors, most pattern classification methods take into account only the global classification error. In this paper we propose to optimize both error types in classification by means of a multiobjective algorithm in which each error type and the network size is an objective of the fitness function. A modified version of the GProp method (optimization and design of multilayer perceptrons) is used, to simultaneously optimize the network size and the type I and II errors. Keywords: multiobjective optimization, multiobjective evolutionary algorithms, Pareto optimality, artificial neural networks optimization, gprop, pattern classification. [ABSTRACT FROM AUTHOR]

This paper presents a technique called Improved Squeaky Wheel Optimisation (ISWO) for driver scheduling problems. It improves the original Squeaky Wheel Optimisation's (SWO) effectiveness and execution speed by incorporating two additional steps of Selection and Mutation which implement evolution within a single solution. In the ISWO, a cycle of Analysis-Selection-Mutation-Prioritization-Construction continues until stopping conditions are reached. The Analysis step first computes the fitness of a current solution to identify troublesome components. The Selection step then discards these troublesome components probabilistically by using the fitness measure, and the Mutation step follows to further discard a small number of components at random. After the above steps, an input solution becomes partial and thus the resulting partial solution needs to be repaired. The repair is carried out by using the Prioritization step to first produce priorities that determine an order by which the following Construction step then schedules the remaining components. Therefore, the optimisation in the ISWO is achieved by solution disruption, iterative improvement and an iterative constructive repair process performed. Encouraging experimental results are reported. [ABSTRACT FROM AUTHOR]

This paper presents two nature-inspired approaches to the design of Zero Reference Codes (ZRC) for optical applications, both in one and two dimensions. Specifically we present a genetic algorithm and a simulated annealing hybridized with a restricted search operator to cope with the problem constraints. Extensive experiments have shown that nature-inspired approaches proposed can improve the results of existing techniques for this problem. [ABSTRACT FROM AUTHOR]

Membrane Computing and Brane Calculi are two recent computational paradigms in the framework of Natural Computing. They are based on the study of the structure and functioning of living cells as living organisms able to process and generate information. In this paper we give a short introduction to both areas and point out some open research lines. [ABSTRACT FROM AUTHOR]

This paper studies the utility of using substructural neighborhoods for local search in the Bayesian optimization algorithm (BOA). The probabilistic model of BOA, which automatically identifies important problem substructures, is used to define the structure of the neighborhoods used in local search. Additionally, a surrogate fitness model is considered to evaluate the improvement of the local search steps. The results show that performing substructural local search in BOA significatively reduces the number of generations necessary to converge to optimal solutions and thus provides substantial speedups. [ABSTRACT FROM AUTHOR]

The paper introduces a hybrid Tabu Search-Evolutionary Algorithm for solving the constraint satisfaction problem, called STLEA. Extensive experimental fine-tuning of parameters of the algorithm was performed to optimise the performance of the algorithm on a commonly used test-set. The performance of the STLEA was then compared to the best known evolutionary algorithm and benchmark deterministic and non-deterministic algorithms. The comparison shows that the STLEA improves on the performance of the best known evolutionary algorithm but can not achieve the efficiency of the deterministic algorithms. [ABSTRACT FROM AUTHOR]

A novel negative selection algorithm, namely r[]-NSA, is proposed in this paper, which uses an array to store multiple partial matching lengths for each detector. Every bit of one detector is assigned a partial matching length. As for a detector, the partial matching length of one bit means that one string is asserted to be matched by the detector, if and only if the number of the maximal continuous identical bits between them from the position of the bit to the end of strings is no less than the partial matching length, and the continuous identical bits should start from the position of the bit. The detector generation algorithm and detection algorithm of r[]-NSA are given. Experimental results showed that r[]-NSA has better detector generation efficiency and detection performance than traditional negative selection algorithm. [ABSTRACT FROM AUTHOR]

In this paper, we propose a variant of an ACO algorithm called the cunning Ant System (cAS). In cAS, each ant generates a solution by borrowing a part of a solution which was generated in previous iterations, instead of generating the solution entirely from pheromone density. Thus we named it, cunning ant. This cunning action reduces premature stagnation and exhibits good performance in the search. The experimental results showed cAS worked very well on the TSP and it may be one of the most promising ACO algorithms. [ABSTRACT FROM AUTHOR]

In this paper we relate information theory and Kolmogorov Complexity (KC) to optimization in the black box scenario. We define the set of all possible decisions an algorithm might make during a run, we associate a function with a probability distribution over this set and define accordingly its entropy. We show that the expected KC of the set (rather than the function) is a better measure of problem difficulty. We analyze the effect of the entropy on the expected KC. Finally, we show, for a restricted scenario, that any permutation closure of a single function, the finest level of granularity for which a No Free Lunch Theorem can hold [7], can be associated with a particular value of entropy. This implies bounds on the expected performance of an algorithm on members of that closure. [ABSTRACT FROM AUTHOR]

This paper presents first results of an analysis of the σ-self-adaptation mechanism on the sharp ridge for non-recombinative (1,λ) evolution strategies (ES). To analyze the ES's evolution, we consider the so-called evolution equations which describe the one-generation change. Neglecting stochastic perturbations and considering only the mean value dynamics, we will investigate possible causes why self-adaptation can fail on the sharp ridge. [ABSTRACT FROM AUTHOR]

In this paper, we show universal lower bounds for isotropic algorithms, that hold for any algorithm such that each new point is the sum of one already visited point plus one random isotropic direction multiplied by any step size (whenever the step size is chosen by an oracle with arbitrarily high computational power). The bound is 1-O(1/d) for the constant in the linear convergence (i.e. the constant C such that the distance to the optimum after n steps is upper bounded by Cn), as already seen for some families of evolution strategies in [19,12], in contrast with 1-O(1) for the reverse case of a random step size and a direction chosen by an oracle with arbitrary high computational power. We then recall that isotropy does not uniquely determine the distribution of a sample on the sphere and show that the convergence rate in isotropic algorithms is improved by using stratified or antithetic isotropy instead of naive isotropy. We show at the end of the paper that beyond the mathematical proof, the result holds on experiments. We conclude that one should use antithetic-isotropy or stratified-isotropy, and never standard-isotropy. [ABSTRACT FROM AUTHOR]

The ridge function class is a parameterised family of test functions that is often used to evaluate the capabilities and limitations of optimisation strategies. Past research with the goal of analytically determining the performance of evolution strategies on the ridge has focused either on the parabolic case or on simple one-parent strategies without step length adaptation. This paper extends that research by studying the performance of multirecombination evolution strategies with cumulative step length adaptation for a wide range of ridge topologies. [ABSTRACT FROM AUTHOR]

This paper presents a particle swarm optimizer to solve constrained optimization problems. The proposed approach adopts a simple method to handle constraints of any type (linear, nonlinear, equality and inequality), and it also presents a novel mechanism to update the velocity and position of each particle. The approach is validated using standard test functions reported in the specialized literature and it's compared with respect to algorithms representative of the state-of-the-art in the area. Our results indicate that the proposed scheme is a promising alternative to solve constrained optimization problems using particle swarm optimization. [ABSTRACT FROM AUTHOR]

This paper extends a geometric framework for interpreting crossover and mutation[5] to the case of sets and related representations. We show that a deep geometric duality exists between the set representation and the vector representation. This duality reveals the equivalence of geometric crossovers for these representations. [ABSTRACT FROM AUTHOR]

For most combinatorial optimization problems the computational complexity of evaluating a single solution is much higher than the cost of evaluating an incremental change made by a local search operator. To benefit from this computational gain crossover can be implemented as a path-following algorithm. As a result crossover becomes more similar to path relinking. In this paper we compare the search bias of crossover and path relinking for permutation problems where the absolute position of the elements is decisive. Calculations show that uniform permutation crossover (UPX) can reach many more permutations from a given parent couple than path relinking. UPX is therefore more exploratory than random path relinking, which is itself more exploratory than greedy path relinking. It is important for users to understand the differences in search bias of the operators so they can choose the exploration operator which they deem most fit for their problem. We conclude with a small experiment on an instance of the quadratic assignment problem. [ABSTRACT FROM AUTHOR]

This paper explores two questions: 1) On a relatively difficult and varied set of test problems, can we observe differences in evolutionary search algorithm performance related to problem features? 2) How do the evolutionary algorithms compare to Pattern Search algorithms, a more traditional optimization tool popular in the larger scientific community? The results suggest there are consistent differences in algorithm performance that can be related back to problem features. Some new ideas for the construction of benchmark problems are also introduced. [ABSTRACT FROM AUTHOR]

Human genetics is undergoing an information explosion. The availability of chip-based technology facilitates the measurement of thousands of DNA sequence variation from across the human genome. The challenge is to sift through these high-dimensional datasets to identify combinations of interacting DNA sequence variations that are predictive of common diseases. The goal of this paper was to develop and evaluate a genetic programming (GP) approach for attribute selection and modeling that uses expert knowledge such as Tuned ReliefF (TuRF) scores during selection to ensure trees with good building blocks are recombined and reproduced. We show here that using expert knowledge to select trees performs as well as a multiobjective fitness function but requires only a tenth of the population size. This study demonstrates that GP may be a useful computational discovery tool in this domain. [ABSTRACT FROM AUTHOR]

In this paper we analyze a new method for an adaptive variation of Evolutionary Algorithms (EAs) population size: the Self-Regulated Population size EA (SRP-EA). An empirical evaluation of the method is provided by comparing the new proposal with the CHC algorithm and other well known EAs with varying population. A fitness landscape generator was chosen to test and compare the algorithms: the Spear's multimodal function generator. The performance of the algorithms was measured in terms of success rate, quality of the solutions and evaluations needed to attain them over a wide range of problem instances. We will show that SRP-EA performs well on these tests and appears to overcome some recurrent drawbacks of traditional EAs which lead them to local optima premature convergence. Also, unlike other methods, SRP-EA seems to self-regulate its population size according to the state of the search. [ABSTRACT FROM AUTHOR]

In this paper we seek an answer to the following question: Is it possible and rewarding to self-adapt parameters regarding selection and population size in an evolutionary algorithm? The motivation comes from the observation that the majority of the existing EC literature is concerned with (self-)adaptation of variation operators, while there are indications that (self-)adapting selection operators or the population size can be equally or even more rewarding. We approach the question in an empirical manner. We design and execute experiments for comparing the performance increase of a benchmark EA when augmented with self-adaptive control of parameters concerning selection and population size in isolation and in combination. With the necessary caveats regarding the test suite and the particular mechanisms used we observe that self-adapting selection yields the highest benefit (up to 30-40%) in terms of speed. [ABSTRACT FROM AUTHOR]

Huge color class redundancy makes the graph coloring problem (GCP) very challenging for genetic algorithms (GAs), and designing effective crossover operators is notoriously difficult. Thus, despite the predominance of population based methods, crossover plays a minor role in many state-of-the-art approaches to solving the GCP. Two main encoding methods have been adopted for heuristic and GA methods: direct encoding, and order based encoding. Although more success has been achieved with direct approaches, algorithms using an order based representation have one powerful advantage: every chromosome decodes as a feasible solution. This paper introduces some new order based crossover variations and demonstrates that they are much more effective on the GCP than other order based crossovers taken from the literature. [ABSTRACT FROM AUTHOR]

This paper analyses the properties of four alternative representation/operator combinations suitable for data clustering algorithms that keep the number of clusters variable. These representations are investigated in the context of their performance when used in a multiobjective evolutionary clustering algorithm (MOCK), which we have described previously. To shed light on the resulting performance differences observed, we consider the relative size of the search space and heuristic bias inherent to each representation, as well as its locality and heritability under the associated variation operators. We find that the representation that performs worst when a random initialization is employed, is nevertheless the best overall performer given the heuristic initialization normally used in MOCK. This suggests there are strong interaction effects between initialization, representation and operators in this problem. [ABSTRACT FROM AUTHOR]

The design of non-cryptographic hash functions by means of evolutionary computation is a relatively new and unexplored problem. In this paper, we use the Genetic Programming paradigm to evolve collision free and fast hash functions. For achieving robustness against collision we use a fitness function based on a non-linearity concept, producing evolved hashes with a good degree of Avalanche Effect. The other main issue, efficiency, is assured by using only very fast operators (both in hardware and software) and by limiting the number of nodes. Using this approach, we have created a new hash function, which we call gp-hash, that is able to outperform a set of five human-generated, widely-used hash functions. [ABSTRACT FROM AUTHOR]

The Maximum Parsimony problem aims at reconstructing a phylogenetic tree from DNA sequences while minimizing the number of evolutionary changes. Known to be NP-complete, the MP problem has many applications. This paper introduces a Distance-based Information Preservation (DiBIP) Tree Crossover. Contrary to previous crossover operators, DiBIP uses a distance measure to characterize the semantic information of a phylogenetic tree and ensures the preservation of distance related properties between parents and offspring. The performance of DiBIP is assessed with a mimetic algorithm on a set of 28 benchmark instances from the literature. Comparisons with 3 state-of-the-art algorithms show very competitive results of the proposed approach with improvement of some previously best results found. [ABSTRACT FROM AUTHOR]

This paper expands Mosaic World, an artificial life model, in order to directly test theories on the emergence of multicellular life. Five experiments are conducted and demonstrate that both the presence of predation and accidental aggregation are sufficient conditions for the transition to multicellularity. In addition, it is shown that division of labour is a major benefit for aggregation, and evolves even if aggregates ‘pay' for abilities they do not use. Analysis of evolved results shows multiple parallels to natural systems, such as differentiation in constituent members of an aggregate, and life-like, complex ecosystems. [ABSTRACT FROM AUTHOR]

Biologically realistic computer simulation of vertebrates is a challenging problem with exciting applications in computer graphics and robotics. Once the mechanics of locomotion are available it is interesting to mediate this locomotion with higher level behavior such as target tracking. One recent approach simulates a relatively simple vertebrate, the lamprey, using recurrent neural networks to model the central pattern generator of the spine and a physical model for the body. Target tracking behavior has also been implemented for such a model. However, previous approaches suffer from deficiencies where particular orientations of the body to the target cause the central pattern generator to shutdown. This paper describes an approach to making target tracking more robust. [ABSTRACT FROM AUTHOR]

In this paper, we present a new parallel self-organizing technique for three dimensional shape reconstruction for mobile robotics. The method is based on adaptive input data decomposition, parallel shape reconstruction in decomposed clusters using Kohonen Self-Organizing Map, which creates mesh representation of the input data. Afterwards, the sub-maps are joined together and the final mesh is re-optimized. Our method overcomes a problem of fitting one mesh to complex non-continuous shapes like building interiors. The method allows to process unordered data collected by mobile robots. The method is easily paralelizable and gives promising results. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of identifying gene modules on the basis of different types of biological data such as gene expression and protein-protein interaction data. Given one or several genes of interest, the aim is to find a group of genes—containing the prespecified genes—that are maximally similar with respect to all data types and sets under consideration. While existing studies follow an aggregation approach to tackle the problem of data integration in module identification, we here propose a multiobjective evolutionary method that provides several advantages: (i) no overall similarity measure needs to be defined, (ii) the interactions and conflicts between the data sets can be explored, and (iii) arbitrary data types can be integrated. The usefulness of the presented approach is demonstrated on different biological scenarios, also in comparison to standard clustering. [ABSTRACT FROM AUTHOR]

The evolution of artificial neural networks (ANNs) is often used to tackle difficult control problems. There are different approaches to the encoding of neural networks in artificial genomes. Analog Genetic Encoding (AGE) is a new implicit method derived from the observation of biological genetic regulatory networks. This paper shows how AGE can be used to simultaneously evolve the topology and the weights of ANNs for complex control systems. AGE is applied to a standard benchmark problem and we show that its performance is equivalent or superior to some of the most powerful algorithms for neuroevolution in the literature. [ABSTRACT FROM AUTHOR]

This paper examines the application of neural networks as a construction heuristic for the examination timetabling problem. Building on the heuristic ordering technique, where events are ordered by decreasing scheduling difficulty, the neural network allows a novel dynamic, multi-criteria approach to be developed. The difficulty of each event to be scheduled is assessed on several characteristics, removing the dependence of an ordering based on a single heuristic. Furthermore, this technique allows the ordering to be reviewed and modified as each event is scheduled; a necessary step since the timetable and constraints are altered as events are placed. Our approach uses a Kohonen self organising neural network and is shown to have wide applicability. Results are presented for a range of examination timetabling problems using standard benchmark datasets. [ABSTRACT FROM AUTHOR]

Diversity maintenance of solutions is an essential part in multi-objective optimization. Existing techniques are suboptimal either in the sense of obtained distribution or execution time. This paper proposes an effective and relatively fast method for pruning a set of non-dominated solutions. The proposed method is based on a crowding estimation technique using nearest neighbors of solutions in Euclidean sense, and a technique for finding these nearest neighbors quickly. The method is experimentally evaluated, and results indicate a good trade-off between the obtained distribution and execution time. Distribution is good also in many-objective problems, when number of objectives is more than two. [ABSTRACT FROM AUTHOR]