Search

The digital age came with an extraordinary ability to generate data across organizations, people, and devices, data that needs to be analyzed, processed and stored. A well-known technique for analyzing this kind of data is Clustering. Many bio-inspired algorithms were proposed for this problem such as the Social Spider Optimization (SSO). In this work, we propose parallel island models of the SSO algorithm for the Clustering problem, using 24 processors for each parallel algorithm. Such models were implemented using static and dynamic topologies, and datasets from the UCI Machine Learning Repository used for the stage of experiments. The achieved average speedups range from 15 to 28 times faster than the SSO algorithm for large and small datasets, respectively, and a parallel model with static ring topology performs a little bit faster than the other parallel models. The parallel algorithms provide results with similar precision to the ones computed with the SSO algorithm.

Parallel island models are used to increase accuracy and performance (speed-up) of meta-heuristics. Such models provide gains by the exchange of information between islands through the migratory process. The key to obtaining gains with parallel island models is the manipulation of migration parameters, since depending on how these parameters are handled the gains vary. Based on this assumption, this work uses three meta-heuristics: genetic algorithm, self-adjusting particle swarm optimization and social spider algorithm. From each metaheuristic, parallel island models were proposed, diversifying the number of natives on the islands, and the behavior of these models were studied. The assessment confirmed the impact of variations migration parameters on accuracy and performance as well as the importance on the number of natives located on the islands. The best solutions were obtained with island models from genetic algorithm and self-adjusting particle swarm optimization, and the best speedups were achieved with island models from social spider algorithm.

The earthquakes occur anywhere and anytime, are one of the most dangerous natural disasters. Magnitude higher or equal to 5 can cause human, material and economic losses. The objective of this paper is association rules search for predict the occurrence of earthquake. To meet the objective was used earthquake data occurred between the years 2000 to 2009, data from tectonic plate near to South America and the Apriori Algorithm. As a result of the study it has 4 association rules of interest were obtained with the probability that one of them occurs, the first pattern interest is “existe_p7_94horas then existe_p7_120horas”, “p7” is the South American Plate.

Designing efficient parallel island Genetic Algorithms (GA) is a difficult task: several decisions are needed related to the adequate structure of the islands, how they are connected, how many individuals should migrate, and how often they should migrate. The impact of these choices has not yet been fully understood since they might vary for different problems. In previous work, a variety of island model GAs to solve Reversal Distance Problem (RDP) over uni-crhomosomal genomes were proposed from which adequate choices were pointed out that provided results with an excellent balance among accuracy and performance. In this work, another evolutionary problem is considered in order to analyze how general were the decisions taken for island model GAs over RDP. The problem is translocation distance over multi-chromosomal genomes, which involves the interchange of gene between different chromosomes. Despite the fact that this problem falls also in the category of evolutionary distance problems, it is different from the RDP. Regarding accuracy, island models using a dynamic communication topology for exchange of individuals between islands provided the best solutions; while regarding performance, models using a static topology reached the highest speedup. Comparing with previous work on RDP, it was observed that islands models that did not provided good accuracy in RDP provided good quality solutions for translocation distance problem, while the best island models for RDP did not repeat the same success for translocation distance problem. The only invariant is that all the island model GAs in addition to competitive speedups provided better results than the corresponding sequential GA.

Computing evolutionary distances using gene order data is a complex combinatory problem; nevertheless, for specific metrics exact polynomial algorithms were proposed, having in many cases non trivial approaches. This scenario can become harder if we want to reconstruct phylogenies based on gene order data: first it is necessary to explore the search space of possible tree structures which is well-known to be exponential; second, it is necessary a method for evaluating the cost of these trees, i.e. to find a labeling of the internal nodes that leads to the most parsimonious cost of a tree under a given evolutionary distance. The latter problem was shown to be NP-hard even for 3 genomes (median problem) under many evolutionary distances. In this paper we propose a variable neighborhood search approach for solving the large phylogeny problem for data based on gene orders. Also, a greedy approach is proposed for the small phylogeny problem aiming to reduce the running time of the Kovac et al. dynamic programming approach. Our proposed algorithms were implemented as the software called HELPHY. Experiments showed that the running time is improved for finding trees with good scores (reversal distance) for the Campanulaceae dataset, and a new tree structure was found having the best known score (double cut and join distance) for the case of Hemiascomycetes dataset.

Rearrangement by reversals is a suitable global operation when treating genomes with a single chromosome. Sorting unsigned genomes by reversals is an NP-hard optimization problem. Several approximation algorithms were proposed, among them, in previous work, a competitive genetic algorithm and its standard parallel version, that provides a substantial speedup, were introduced. In this paper, two approaches using island models to parallelize such algorithm are presented. The first approach uses the unidirectional ring communication topology to exchange individuals between neighboring islands and, the second uses a complete graph scheme for the distribution of individuals among islands. Both approaches were proposed with the objective of improving precision (that is, for reducing the number of reversals) and decreasing the runtime regarding the sequential GA. Experiments were performed with randomly generated synthetic genomes and the results show that the parallel approach using the ring communication topology outperforms the previously proposed GA and its parallel version in terms of accuracy, providing solutions with less reversals and, that the parallel approach using the complete graph topology does not provide significant improvements. Both the new parallel GA approaches get competitive speedups regarding the speedup achieved by the standard parallel version of the genetic algorithm.

The rearrangement of genomes is an important tool for studying the evolution of genomes and specifically for the construction of phylogenies. A translocation splits and combines the strings of genes of a pair of chromosomes inside a genome and is considered a suitable operation for rearrangement of genomes with multiple chromosomes. The translocation distance between two genomes is the minimum number of translocations necessary to convert one of them into the other. Computing the translocation distance between two unsigned genomes, that is the case in which the direction of the genes between the chromosomes is not considered, is known to be an MV-hard optimization problem. Among several approximation algorithms that were proposed for solving this problem, the authors introduced in a previous work a genetic algorithm approach improved with opposition based learning and memetic mechanisms. In this paper, two parallel treatments of the sequential memetic approach are introduced for solving the translocation distance problem for unsigned genomes. The first approach, computes in parallel the fitness over all individuals of a population. This method intends speeding-up the sequential memetic algorithm. The second approach, processes in parallel multiple populations and was proposed for improving precision providing solutions with a less number of translocations than the sequential memetic algorithm. Several experiments were performed with randomly generated synthetic and biologically based genomes. Results show that the parallel approaches outperform the sequential memetic algorithm.

A standard genetic algorithm (\(\mathcal{G\!A}_{{ \mathrm{S}}}\)) for sorting unsigned genomes by translocations is improved in two different manners: firstly, a memetic algorithm (\(\mathcal{G\!A}_{{ \mathrm{M}}}\)) is provided, which embeds a new stage of local search, based on the concept of mutation applied in only one gene; secondly, an opposition-based learning (\(\mathcal{G\!A}_{{ \mathrm{OBL}}}\)) mechanism is provided that explores the concept of internal opposition applied to a chromosome. Both approaches include a convergence control mechanism of the population using the Shannon entropy. For the experiments, both biological and synthetic genomes were used. The results showed that \(\mathcal{G\!A}_{{ \mathrm{M}}}\)outperforms both \(\mathcal{G\!A}_{{ \mathrm{S}}}\)and \(\mathcal{G\!A}_{{ \mathrm{OBL}}}\)as confirmed through statistical tests.

Translocation is a useful operation on strings with challenging questions in combinatorics of permutations and interesting applications in analysis of sequences. A translocation operation essentially is the interchange of prefixes and suffixes among two substrings of a string. For the case of genomes represented as strings, symbols that represent genes and chromosomes are modeled as substrings of the genomes; thus, translocation is an operation that models the interaction between chromosomes inside a genome. The translocation distance between two genomes is defined as the minimum number of translocations to convert one genome into another and has been proved to be a meaningful manner of modeling the evolutive distance between organisms. The particular case of unsigned genomes, those in which the orientation of the genes are not considered, is particularly difficult, while the signed case, in which the orientation of genes is considered, has been proved to be polynomially decidable. This paper presents an innovative Genetic Algorithm (GA) approach to solve the unsigned translocation distance problem. A distinguishing feature of the proposed GA is that it uses as fitness function the translocation distance for randomly generated signed versions of the input (that is an unsigned genome). Experiments over randomly generated strings (synthetic genomes) showed that the proposed GA approach computes answers that are better than those computed by an L5+e-approximation algorithm, the latter also implemented as part of this work.

Reversals are operations of great biological significance for the analysis of the evolutionary distance between organisms. Genome rearrangement through reversals, consists in finding the shortest sequence of reversals to transform one genome represented as a signed or unsigned permutation into another. When genes are non oriented and correspondingly permutations are unsigned, sorting by reversals came arise as a challenging problem in combinatorics of permutations. In fact, this problem is known to be NP-hard, but the question whether it is NP-complete remains open for more than twenty years. When permutations are signed and correspondingly genes are oriented, the problem is known to be in P. A parallelization of a standard GA (Genetic Algorithm) is proposed for the problem of sorting unsigned permutations. This GA was previously reported in the literature as the most competitive regarding precision for which as control mechanism an 1.5-approximation algorithm was used. For the parallelization, the MPI Library of the C language was used and experiments were performed for calculating the execution time and precision. By increasing the number of individuals, experiment showed improvement in relation to previous approaches. Additionally, a virtualization of the GA using a MicroBlaze processor from Xilinx was performed on OVP for which the average number of executed instructions was of approximately 1.40 Giga instruction per second.

Sorting unsigned permutations by reversals is an important and difficult problem in combinatorial processing of permutations with important applications in bio-informatics for the interpretation of the evolutionary relationship between organisms. Since it was shown that the problem is NP-hard many approximation and a few evolutionary algorithms were proposed. In this paper we propose a new genetic algorithm approach that uses modified crossover and mutation operators adapted to the problem. Instead previous genetic algorithmic approaches, the proposed algorithm uses a very simple fitness function that can be linearly computed in the size of the permutation and updated in constant time, for each individual in each generation. In order to compare the accuracy of the computed solutions, an 1.5 approximation ratio algorithm was developed by fixing Christie's approximation method. The results showed that on average the proposed genetic approach produces competitive results in relation with the ones given by the 1.5-approximation algorithm. Additionally, it has been observed that for permutations of all sizes, that were randomly generated, it was alway possible to compute better solutions with the genetic than with the approximate approach and that the difference obtained for these cases is greater than the ones obtained in the cases in which the genetic approach has a worse behavior than the approximate one.