In recent decades, the explosive growth of spatial data has presented significant challenges in managing complex datasets and creating meaningful thematic maps. The flexibility and adaptability of evolutionary computing provide powerful solutions that are particularly advantageous for addressing these multifaceted spatial analysis tasks. By emulating the process of natural selection, evolutionary algorithms offer a flexible mechanism to explore complex solution spaces and provide interpretable, high-quality results without excessive dependence on manual parameter tuning. In this work, we introduce GIS-moGA, a multi-objective genetic algorithm designed to optimize the composition of thematic maps by maximizing spatial coherence. Traditional Geographic Information Systems (GIS) rely on expert-driven weight assignments, which can introduce subjectivity and limit reproducibility. Our approach employs spatial statistics, including Global Moran’s I and Local Indicators of Spatial Association , to generate high-quality, data-driven thematic maps for epidemiological analysis. We validate GIS-moGA through a case study in São Carlos, Brazil, integrating demographic, socioeconomic, and epidemiological datasets to identify high-risk areas for diseases such as dengue, COVID-19, and tuberculosis. The results demonstrate that the proposed method effectively reduces reliance on manual parameter tuning while improving spatial coherence and decision support capabilities. Despite its advantages, GIS-moGA has limitations, including computational scalability challenges and reliance on predefined spatial statistical measures. Future work aims to enhance the algorithm’s efficiency, incorporate real-time data streams, and extend its applicability to broader domains such as disaster response and climate resilience. In essence, this research provides a robust, scalable, and interpretable approach to spatial data optimization, ultimately improving public health interventions and resource allocation.

Extracting meaningful insights from tabular data remains a fundamental challenge in machine learning, as traditional methods often struggle to capture complex feature interactions. In this work, we propose a novel graph-based approach for analyzing tabular datasets by leveraging spectral graph theory and community detection techniques. Our method represents tabular data as a weighted directed graph, where edges encode feature dependencies based on SHAP values. To enhance interpretability, we apply a sparsification technique that retains only the most significant connections. We further analyze the structural properties of the resulting graph using the deformed magnetic Laplacian, which captures directional dependencies among features. Additionally, we employ a nonparametric stochastic block model (nSBM) to uncover hierarchical modular structures and use tabular embeddings (tab2vec) to reveal fine-grained relationships in feature space. Our framework is validated on the PeNSE dataset, a large-scale survey on adolescent health, demonstrating its ability to reveal hidden structures and improve feature interpretability. Results show that spectral analysis provides an effective way to categorize features into meaningful clusters, identify redundant variables, and highlight key relationships that may be overlooked by conventional techniques. This approach offers a powerful alternative for exploring complex tabular datasets, with potential applications in various domains such as healthcare, finance, and social sciences.

Although it is acknowledged that urban inequalities can lead to biases in the production of social media data, there is a lack of studies which make an assessment of the effects of intra-urban movements in real-world urban analytics applications, based on social media. This study investigates the spatial heterogeneity of social media with regard to the regular intra-urban movements of residents by means of a case study of rainfall-related Twitter activity in São Paulo, Brazil. We apply a spatial autoregressive model that uses population and income as covariates and intra-urban mobility flows as spatial weights to explain the spatial distribution of the social response to rainfall events in Twitter vis-à-vis rainfall radar data. Results show high spatial heterogeneity in the response of social media to rainfall events, which is linked to intra-urban inequalities. Our model performance (R2=0.80) provides evidence that urban mobility flows and socio-economic indicators are significant factors to explain the spatial heterogeneity of thematic spatiotemporal patterns extracted from social media. Therefore, urban analytics research and practice should consider not only the influence of socio-economic profile of neighborhoods but also the spatial interaction introduced by intra-urban mobility flows to account for spatial heterogeneity when using social media data.

The spatial analysis of social media data has recently emerged as a significant source of knowledge for urban studies. Most of these analyses are based on an areal unit that is chosen without the support of clear criteria to ensure representativeness with regard to an observed phenomenon. Nonetheless, the results and conclusions that can be drawn from a social media analysis to a great extent depend on the areal unit chosen, since they are faced with the well-known Modifiable Areal Unit Problem. To address this problem, this article adopts a data-driven approach to determine the most suitable areal unit for the analysis of social media data. Our multicriteria optimization framework relies on the Pareto optimality to assess candidate areal units based on a set of user-defined criteria. We examine a case study that is used to investigate rainfall-related tweets and to determine the areal units that optimize spatial autocorrelation patterns through the combined use of indicators of global spatial autocorrelation and the variance of local spatial autocorrelation. The results show that the optimal areal units (30 km2 and 50 km2) provide more consistent spatial patterns than the other areal units and are thus likely to produce more reliable analytical results.

Summary Science Gateways provide portals for experiments execution, regardless of the users' computational background. Nowadays its construction and performance need enhancement in terms of resource provision and task scheduling. We present the Modular Distributed Architecture to support the Protein Structure Prediction (MDAPSP), a Service‐Oriented Architecture for management and construction of Science Gateways, with resource provisioning on a heterogeneous environment. The Decision Maker, central module of MDAPSP, defines the best computational environment according to experiment parameters. The proof of concept for MDAPSP is presented in WorkflowSim, with two novel schedulers. Our results demonstrate good Quality of Service (QoS), capable of correctly distributing the workload, fair response times, providing load balance, and overall system improvement. The study case relies on PSP algorithms and the Galaxy framework, with monitoring experiments to show the bottlenecks and critical aspects.

Optimization methods for spanning tree problems may require efficient data structures. The node-depth-degree representation (NDDR) has achieved relevant results for direct spanning tree representation together with evolutionary algorithms (EAs). Its two mutation operators have average time O(√n), where n is the number of vertices of the graph, while similar operators implemented by predecessor arrays, a typical tree data structure, have time O(n). Dynamic trees are also relevant when investigating tree representations since they have low time complexity, but there is no proper extension of them for EAs. Using aspects of both a dynamic tree and NDDR, namely, Euler tours and structural sharing, we propose a data structure called 2LETT, whose mutation operators have time O(√n) in the worst case. Experiments with the mutation operators using 2LETT, predecessor arrays, and NDDR are carried out for graphs with up to 300000 vertices. For a mutation operator that exchanges any two valid edges, predecessor arrays present the best performance for random trees with fewer than 10000 vertices; while 2LETT has the best performance for trees with more than 10000 vertices. Especially, noteworthy is the fact that 2LETT is the only structure whose running time is independent of tree diameter.

Every day more and more objects are connected to the Internet to sense or actuate in some environment, composing the Internet of Things. IoT platforms will play a key role, as they will be responsible for managing low-level devices and data acquisition processes, and also support the development of new applications. One of the main challenges in IoT platforms will be the search and discovery of resources in large-scale and heterogeneous environments for reuse by other applications to support their specific requirements. In this paper, we propose an elimination-selection algorithm for search and discovery of resources in IoT environments. Our case study considers a real agricultural problem to be solved by the ViSIoT tool. The results show that our approach improves the quality of the proposed solution adding a small time overhead when compared to the TOPSIS algorithm used by ViSIoT.

Summary The growth of real‐world objects with embedded and globally networked sensors allows to consolidate the Internet of things paradigm and increase the number of applications in the domains of ubiquitous and context‐aware computing. The merging between cloud computing and Internet of things named cloud of things will be the key to handle thousands of sensors and their data. One of the main challenges in the cloud of things is context‐aware sensor search and selection. Typically, sensors require to be searched using two or more conflicting context properties. Most of the existing work uses some kind of multi‐criteria decision analysis to perform the sensor search and selection, but does not show any concern for the quality of the selection presented by these methods. In this paper, we analyse the behaviour of the SAW, TOPSIS and VIKOR multi‐objective decision methods and their quality of selection comparing them with thePareto‐optimality solutions. The gathered results allow to analyse and compare these algorithms regarding their behaviour, the number of optimal solutions and redundancy. Copyright © 2016 John Wiley & Sons, Ltd.

A large number of compiler optimizations are nowadays available to users. These optimizations interact with each other and with the input code in several and complex ways. The sequence of application of optimization passes can have a significant impact on the performance achieved. The effect of the optimizations is both platform and application dependent. The exhaustive exploration of all viable sequences of compiler optimizations for a given code fragment is not feasible. As this exploration is a complex and time-consuming task, several researchers have focused on Design Space Exploration (DSE) strategies both to select optimization sequences to improve the performance of each function of the application and to reduce the exploration time. In this article, we present a DSE scheme based on a clustering approach for grouping functions with similarities and exploration of a reduced search space resulting from the combination of optimizations previously suggested for the functions in each group. The identification of similarities between functions uses a data mining method that is applied to a symbolic code representation. The data mining process combines three algorithms to generate clusters: the Normalized Compression Distance, the Neighbor Joining, and a new ambiguity-based clustering algorithm. Our experiments for evaluating the effectiveness of the proposed approach address the exploration of optimization sequences in the context of the ReflectC compiler, considering 49 compilation passes while targeting a Xilinx MicroBlaze processor, and aiming at performance improvements for 51 functions and four applications. Experimental results reveal that the use of our clustering-based DSE approach achieves a significant reduction in the total exploration time of the search space (20× over a Genetic Algorithm approach) at the same time that considerable performance speedups (41% over the baseline) were obtained using the optimized codes. Additional experiments were performed considering the LLVM compiler, considering 124 compilation passes, and targeting a LEON3 processor. The results show that our approach achieved geometric mean speedups of 1.49 × , 1.32 × , and 1.24 × for the best 10, 20, and 30 functions, respectively, and a global improvement of 7% over the performance obtained when compiling with -O2.

Structured evolutionary algorithms have been investigated for some time. However, they have been under explored especially in the field of multi-objective optimization. Despite good results, the use of complex dynamics and structures keep the understanding and adoption rate of structured evolutionary algorithms low. Here, we propose a general subpopulation framework that has the capability of integrating optimization algorithms without restrictions as well as aiding the design of structured algorithms. The proposed framework is capable of generalizing most of the structured evolutionary algorithms, such as cellular algorithms, island models, spatial predator-prey, and restricted mating based algorithms. Moreover, we propose two algorithms based on the general subpopulation framework, demonstrating that with the simple addition of a number of single-objective differential evolution algorithms for each objective, the results improve greatly, even when the combined algorithms behave poorly when evaluated alone at the tests. Most importantly, the comparison between the subpopulation algorithms and their related panmictic algorithms suggests that the competition between different strategies inside one population can have deleterious consequences for an algorithm and reveals a strong benefit of using the subpopulation framework.

In this paper we present a clustering-based selection approach for reducing the number of compilation passes used in search space during the exploration of optimizations aiming at increasing the performance of a given function and/or code fragment. The basic idea is to identify similarities among functions and to use the passes previously explored each time a new function is being compiled. This subset of compiler optimizations is then used by a Design Space Exploration (DSE) process. The identification of similarities is obtained by a data mining method which is applied to a symbolic code representation that translates the main structures of the source code to a sequence of symbols based on transformation rules. Experiments were performed for evaluating the effectiveness of the proposed approach. The selection of compiler optimization sequences considering a set of 49 compilation passes and targeting a Xilinx MicroBlaze processor was performed aiming at latency improvements for 41 functions from Texas Instruments benchmarks. The results reveal that the passes selection based on our clustering method achieves a significant gain on execution time over the full search space still achieving important performance speedups.

Estimation of Distribution Algorithms (EDAs) have emerged from the synergy between machine-learning techniques and Genetic Algorithms (GAs). EDAs rely on probabilistic modeling for obtaining information about the underlying structure of optimization problems and implementing effective reproduction operators. The effectiveness of EDAs depends on the capacity of the model-building to extract reliable information about the problem. In this study we analyze additively separable functions and argue that the degree of multimodality of such functions defines their linkage-learning difficulty. Besides, by using entropy-based concepts and Jensen's inequality, we show how allelic pairwise independence may appear as a consequence of an increasing multimodality. The results characterize the linkage-learning difficulty of well-known functions, like the deceptive trap, bipolar and concatenated parity.

A new method to group the most effective preventive controls for a set of critical contingencies and select the promising groups considering cost and effectiveness is proposed. This methodology is based on a sensitivity analysis of the maximum loadability point with respect to voltage controls and on a hierarchical clustering analysis. Considering not only the effectiveness of the control elements but also their availability, the methodology design a set of controllers to eliminate all critical contingencies. The methodology was successfully tested in a reduced south-southeast Brazilian system composed of 107 buses.

Due to the large number of optimizations provided in modern compilers and to compiler optimization specific opportunities, a Design Space Exploration (DSE) is necessary to search for the best sequence of compiler optimizations for a given code fragment (e.g., function). As this exploration is a complex and time consuming task, in this paper we present DSE strategies to select optimization sequences to both improve the performance of each function and reduce the exploration time. The DSE is based on a clustering approach which groups functions with similarities and then explore the reduced search space provided by the optimizations previously suggested for the functions in each group. The identification of similarities between functions uses a data mining method which is applied to a symbolic code representation of the source code. The DSE process uses the reduced set identified by clustering in two ways: as the design space or as the initial configuration. In both ways, the adoption of a pre-selection based on clustering allows the use of simple and fast DSE algorithms. Our experiments for evaluating the effectiveness of the proposed approach address the exploration of compiler optimization sequences considering 49 compilation passes and targeting a Xilinx MicroBlaze processor, and were performed aiming performance improvements for 41 functions. Experimental results reveal that the use of our new clustering-based DSE approach achieved a significant reduction on the total exploration time of the search space (18x over a Genetic Algorithm approach for DSE) at the same time that important performance speedups (43% over the baseline) were obtained by the optimized codes.

This paper presents a multi-objective formulation for variable selection in calibration problems. The prediction of protein concentration on wheat is obtained by a linear regression model using variables obtained by a spectrophotometer device. This device measure hundreds of correlated variables related with physicochemical properties and that can be used to estimate the protein concentration. The problem is the selection of a subset informative and uncorrelated variables that help the minimization of prediction error. In this work we propose the use of two objectives in this problem: the prediction error and the number of variables in the model, both related to linear equations system stability. We proposed a multi-objective formulation using two multi-objective algorithms: the NSGA-II and the SPEA-II. Additionally we propose a final decision maker method to choice the final subset of variables from the Pareto front. For the case study is used wheat data obtained by NIR spectrometry where the objective is the determination of a variable subgroup with information about protein concentration. The results of traditional techniques of multivariate calibration as the Successive Projections Algorithm (SPA), Partial Least Square (PLS) and mono-objective genetic algorithm are presents for comparisons. For NIR spectral analysis of protein concentration on wheat, the number of variables selected from 775 spectral variables was reduced for just 10 in the SPEA-II algorithm. The prediction error decreased from 0.2 in the classical methods to 0.09 in proposed approach, a reduction of 45%. The model using variables selected by SPEA-II had better prediction performance than classical algorithms and full-spectrum partial least-squares (PLS).

Linkage Learning (LL) was proposed as a methodology to enable Genetic Algorithms (GAs) to solve complex optimization problems more effectively. Its main idea relies on a reductionist assumption, considering optimization problems as being composed of substructures that could be exploited to improve the GA's search mechanism. In general, LL-GAs have been compared in a restricted set of well-known optimization problems, in which the reductionist assumption holds true, and only a few studies have concerned their performances in broader scenarios. To help to fill this gap, we have compared four different LL-GAs in the classic Multidimensional Knapsack Problem (MKP) using all the instances provided by Chu & Beasley (1998). Our objective was to verify if the relative performance of algorithms as: the Extended Compact Genetic Algorithm (eCGA), the Bayesian Optimization Algorithm (BOA) with decision graphs, the BOA with community detection, the Linkage Tree Genetic Algorithm (LTGA) and a simple GA; would remain the same in the MKP's instances, where the existence of substructures is unknown. However, the results have shown the opposite, and algorithms as BOA have only found similar solutions to those found by the eCGA and LTGA when using large population sizes.

Linkage Learning (LL) is an important issue concerning the development of more effective genetic algorithms (GA). It is from the identification of strongly dependent variables that crossover can be effective and an efficient search can be implemented. In the last decade many algorithms have confirmed the beneficial influence of LL when solving nearly decomposable problems. As it is a well-known fact from the no free-lunch theorem, LL can not be the best tool for all optimization problems, therefore, methods to identify those problems which could be efficiently solved by LL have become necessary. This paper investigates that nearly-decomposable problems present characteristic linkage-trees, therefore, those trees can be used as reference to infer whether or not some black-box optimization problem is a good candidate to be solved by LL. In this context, we consider the linkage-tree model from the Linkage-Tree GA (LTGA) and use the silhouette measure to expose some problems' characteristics. The silhouette fingerprints (SF) are defined for overlapping deceptive trap functions and compared with the SFs obtained for Multidimensional Knapsack Problems (MKP). The comparison allowed us to conclude that MKPs do not present evident linkages. This result was confirmed by experiments comparing the performance of the LTGA and the Randomized LTGA, in which both algorithms had very similar results.

The network reconfiguration for service restoration (SR) in distribution systems is a combinatorial complex optimization problem since it involves multiple non-linear constraints and objectives. For large networks, no exact algorithm has found adequate SR plans in real-time. On the other hand, methods combining Multi-objective Evolutionary Algorithms (MOEAs) with the Node-depth encoding (NDE) have shown to be able to efficiently generate adequate SR plans for large distribution systems (with thousands of buses and switches). This paper presents a new method that combining NDE with three MOEAs: (i) NSGA-II; (iii) SPEA 2; and (iii) a MOEA based on subpopulation tables. The idea is to obtain a method that cannot-only obtain adequate SR plans for large scale distribution systems, but can also find plans for small or large networks with similar quality. The proposed method, called MEA2N-STR, explores the space of the objectives solutions better than the other MOEAs with NDE, approximating better the Pareto-optimal front. This statement has been demonstrated by several simulations with DSs ranging from 632 to 1,277 switches.

A new methodology for grouping and selecting the most effective controls for a group of critical contingencies to prevent voltage instability in electrical power systems is developed in this paper. This methodology is based on a sensitivity analysis of the maximum loadability point with respect to voltage controls and on a hierarchical clustering analysis. Considering not only the effectiveness of the control elements but also their availability, the methodology design a set of controllers to eliminate all critical contingencies. The methodology is fast and suitable for on-line assessment of preventive control. The methodology was successfully tested in a reduced south-southeast Brazilian system composed of 107 buses.

SUMMARY Severalab initiocomputational methods for protein structure prediction have been designed using full‐atom models and force field potentials to describe interactions among atoms. Those methods involve the solution of a combinatorial problem with a huge search space. Genetic algorithms (GAs) have shown significant performance increases for such methods. However, even a small protein may require hundreds of thousands of energy function evaluations making GAs suitable only for the prediction of very small proteins. We propose an efficient technique to compute the van der Waals energy (the greatest contributor to protein stability) speeding up the whole GA. First, we developed a Cell‐List Reconstruction procedure that divides the tridimensional space into a cell grid for each new structure that the GA generates. The cells restrict the calculations of van der Waals potentials to ranges in which they are significant, reducing the complexity of such calculations from quadratic to linear. Moreover, the proposal also uses the structure of the cell grid to parallelize the computation of the van der Waals energy, achieving additional speedup. The results have shown a significant reduction in the run time required by a GA. For example, the run time for the prediction of a protein with 147,980 atoms can be reduced from 217 days to 7 h. Copyright © 2012 John Wiley & Sons, Ltd.

The Protein Structure Prediction (PSP) problem is concerned about the prediction of the native structure of a protein from its amino acid sequence. PSP is a challenging and computationally open problem. Therefore, several researches and methodologies have been developed for it. This paper presents the application of protpred-GROMACS, an evolutionary framework for PSP, in structural and energetic contexts. The performance of mono-objective algorithm was compared with other methodologies, such as multi-objective evolutionary algorithm, coarse grained monte carlo and replica exchange molecular dynamics.

The design of a network is a solution to several engineering and science problems. Several network design problems are known to be NP-hard, and population-based metaheuristics like evolutionary algorithms (EAs) have been largely investigated for such problems. Such optimization methods simultaneously generate a large number of potential solutions to investigate the search space in breadth and, consequently, to avoid local optima. Obtaining a potential solution usually involves the construction and maintenance of several spanning trees, or more generally, spanning forests. To efficiently explore the search space, special data structures have been developed to provide operations that manipulate a set of spanning trees (population). For a tree withnnodes, the most efficient data structures available in the literature require timeO(n) to generate a new spanning tree that modifies an existing one and to store the new solution. We propose a new data structure, called node-depth-degree representation (NDDR), and we demonstrate that using this encoding, generating a new spanning forest requires average timeO(√n). Experiments with an EA based on NDDR applied to large-scale instances of the degree-constrained minimum spanning tree problem have shown that the implementation adds small constants and lower order terms to the theoretical bound.

Several computational models have been developed in the context of the Protein Structure Prediction (PSP) problem. These methods involve a combinatorial problem and can be solved using optimizing algorithms in order to search for a global minimum energy. Genetic Algorithms (GAs) have produced relevant results in this area. Several energies in the protein are known to be directly responsible for the stabilization of their structures. These energies can represent each objective of multiobjective evolutionary algorithms. Many techniques, as the NSGA-II, are used to deal with the multi-objective approach for proteins, however they are not adequate for the PSP problem. New strategies have been sought with multiple criteria. In this context, this paper introduces the application of multiobjective evolutionary algorithm on tables algorithm to the PSP problem. In order to evaluate this approach, we compare it with the well-known NSGA-II algorithm. The new approach investigated for PSP can generate protein structures with energies significantly smaller than those generated by the NSGA-II.

Van der Waals's energy models attraction and repulsion effects between pairs of atoms. This energy is used by ab initio methods to find the tertiary structure of a protein based only on its amino acid sequence and on a force field model. Several researches suggests Genetic Algorithms (GAs), are adequate for the development of ab initio approaches for protein structure prediction. A GA generates thousands of potential structures for a protein conformation, and evaluates the van der Waals' interaction in each generated structure. In practice, 99% of running time of the GA is used with the computation of van der Waals' energy. To compute the van der Waals energy for a given structure, we need to calculate effects of the interactions of all pairs of atoms in the structure. Using this cutoff, the complexity of the algorithm is O(n2) per conformation, where n is the number of atoms of the protein. For atoms separated by more than 8 Å the van der Waals effect is relatively weak. Thus, we apply a Cell-lists method to the van der Waals function reducing the complexity of algorithm to O(n). Furthermore, we applied parallel programming to the Cell-lists method using MPI, reducing significatively the running time. The combination of the Cell-lists and MPI techniques resulted in a speedup of 1000 for a protein with 147,900 atoms.

In machine learning accurate predictors may be obtained by combining predictions of an ensemble of accurate and diverse predictors. Ensembles are efficiently constructed with the random subspace method (RSM) performed in the instance or in the principal components (PCs) spaces. In this paper, we extend RSM to explore the synergy in the characteristics of these two spaces, with a method referred to as RSM-IPCS. Using 24 datasets from the UCI machine learning repository, we show an enhanced performance of RSM-IPCS in comparison to the original RSM and RSM in PCs space, in terms of higher accuracy and smaller variances. Since RSM-IPCS exhibited at least a similar performance to the best method in a separate space, it opens the way for optimization of ensembles based on the combination of multiple spaces.

This paper presents a novel method to discover promising regions in a continuous search space. Using machine learning techniques, the algorithm named smart sampling was tested in hard known benchmark functions, and was able to find promising regions with solutions very close to the global optimum, significantly decreasing the number of evaluations needed by a metaheuristic to finally find this global optimum, when heuristically started inside a promising region. Results show favorable agreement with theories which state the importance of an adequate starting population. The results also present significant improvement in the efficiency of the tested metaheuristic, without adding any parameter, operator or strategy. Being a technique which can be used by any populational metaheuristic, the work presented here has profound implications for future studies of global optimization and may help solve considerably difficult optimization problems.

This paper describes and analyzes population size management, which can be used to enhance the efficiency of the extended compact genetic algorithm (ECGA). The ECGA is a selectorecombinative algorithm that requires an adequate sampling to generate a high-quality model of the problem. Population size management decreases the overall running time of the optimization process by splitting the algorithm into two phases: first, it builds a high-quality model of the problem using a large population; second, it generates a smaller population, sampled using the high-quality model, and performs the remaining of the optimization with a reduced population size. The paper shows that for decomposable optimization problems, population size management leads to a significant optimization speedup that decreases the number of evaluations for convergence in ECGA by a factor of 30% to 70% keeping the same accuracy and reliability. Furthermore, the ECGA using PSM presents the same scalability model as the ECGA.

This paper evaluates the performance of three evolutionary algorithms to globally optimize complex continuous functions. The performance is evaluated by measuring the algorithms success rate to find the global optimum in several trials. At each set of trials, the search-space is reduced to be closer to the global optimum, so that the starting population is generated in an even more promising region. According to the results, it is possible to can conclude that, in high complexity problems, a good performance of classical evolutionary algorithms can not be expected. The paper also evaluates the performance of an evolutionary algorithm in a deceptive function. In this case, the reduced search-space is the model which generates the deceptive function. The success rates with and without the use of the starting model were compared. In this case, the use of a better starting model substantially increases the performance.

The node-depth encoding has elements from direct and indirect encoding for trees which encodes trees by storing the depth of nodes in a list. Node-depth encoding applies specific search operators that is a typical characteristic for direct encodings. An investigation into the bias of the initialization process and the mutation operators of the node-depth encoding shows that the initialization process has a bias to solutions with small depths and diameters, and a bias towards stars. This investigation, also, shows that the mutation operators are unbiased. The performance of node-depth encoding is investigated for the bounded-diameter minimum spanning tree problem. The results are presented for Euclidean instances presented in the literature. In contrast with the expectation, the evolutionary algorithm using the biased initialization operator does not allow evolutionary algorithms to find better solutions compared to an unbiased initialization. In comparison to other evolutionary algorithms for the bounded-diameter minimum spanning tree evolutionary algorithms using the node-depth encoding have a good performance.

Proteins are polymers whose chains are composed of 20 different monomers, called amino acids. The problem of Protein Structure Prediction (PSP) is the determination of protein 3D conformation from its amino acid sequence. Two main strategies are usually employed to work with PSP: homology and Ab initio approaches. This paper presents an Evolutionary Algorithm to PSP using an Ab initio approach (ProtPred). The predictions are evaluated using fitness functions based on potential energies (electrostatic and van der Waals) and hydrophobic interactions. The proposed approach uses dihedral angles and main angles of the lateral chains to model a protein structure. ProtPred is evaluated using relatively complex cases for an Ab initio approach. Results have shown that ProtPred is a consistent approach.

Various phylogenetic reconstruction methods have been proposed in order to determine the most accurate tree that represents evolutionary relationships among species. Each method defines a criterion for evaluation of possible solutions. This criterion leads the search to the best phylogenetic tree. However, different criteria may lead to distinct phylogenies, which often conflict with each other. In this context, a multi-objective approach can be useful since it could produce a set of optimal trees (Pareto front) according to multiple criteria. We propose a multi-objective evolutionary algorithm, called Phylo-MOEA, which is focused on maximum parsimony and maximum likelihood criteria. In experiments, several PhyloMOEA trials were performed using four datasets of nucleotide sequences. For each dataset, the proposed algorithm found a Pareto front representing a trade-off between the criteria used. Moreover, SH-test showed that a number of solutions from PhyloMOEA are not significantly worse than solutions found by phylogenetic programs using one criterion

The Protein Structure Prediction (PSP) is to determine the proteintertiary structure from its amino acids. This paper presents the ProtPred and investigates its application. The first results showed that ProtPred is a consistent approach.

We have developed an algorithm for reduction of search-space, called Domain Optimization Algorithm (DOA), applied to global optimization. This approach can efficiently eliminate search-space regions with low probability of containing a global optimum. DOA basically worksusing simple models for search-space regions to identify and eliminate non-promising regions. The proposed approach has shown relevant results for tests using hard benchmark functions.

This paper presents a performance comparison of various methods for associative memory design using cellular neural networks (CNNs). Even though there have been an increasing interest in such kind of application for CNNs, there is no proper comparison of performance among the available methods in the literature. This paper reviews methods of associative memory design based on CNNs, and provides comparative performance analyses of these approaches.

Evolutionary algorithms (EAs) are a largely used search and optimization technique. They have been successfully applied to a wide variety of problems, overcoming traditional algorithms in performance. However, few EAs and traditional algorithms are able to handle complex combinatorial problems involving a large number of variables (thousands or millions). This paper proposes a new EA, capable of solving combinatorial problems with large number of variables. This algorithm is the result of two extensions from the extended compact genetic algorithm, a state-of-the-art EA.

There is an increasing concern with the quality of beverage products, like water, milk, wine and coffee. For wine, in particular, the evaluation is usually performed by human tasters, which may require a long time for their training. Moreover, the use of human tasters usually presents high financial and health costs and has a strong subjective component. The quality control of beverages may strongly benefit from the automatic monitoring of their properties by using taste sensors and intelligent systems. This work investigates how the application of a artificial intelligence, more specifically a hybrid case-based system, can lead to an efficient tool for wine quality monitoring. For such, a set of measures extracted by a set of taste sensors is analysed by an intelligent hybrid system. Experimental results show the ability of the proposed system to evaluate wine quality.

Network design involves several areas of research. Computer networks, electrical circuits and transportation problems are some examples. In order to deal with the complexity of these problems, approaches using evolutionary algorithms have been proposed for network design problems (NDPs) with relevant results. Nevertheless, the graph encoding is critical for the performance of evolutionary algorithms for NDPs. The node-depth encoding (NDE) has presented relevant results for NDPs involving undirected graphs. In this sense, this article proposes an extension of NDE for NDPs modeled by directed graphs.