Showing papers on "Selection (genetic algorithm) published in 2003"

Benchmarking attribute selection techniques for discrete class data mining

Abstract: Data engineering is generally considered to be a central issue in the development of data mining applications. The success of many learning schemes, in their attempts to construct models of data, hinges on the reliable identification of a small set of highly predictive attributes. The inclusion of irrelevant, redundant, and noisy attributes in the model building process phase can result in poor predictive performance and increased computation. Attribute selection generally involves a combination of search and attribute utility estimation plus evaluation with respect to specific learning schemes. This leads to a large number of possible permutations and has led to a situation where very few benchmark studies have been conducted. This paper presents a benchmark comparison of several attribute selection methods for supervised classification. All the methods produce an attribute ranking, a useful devise for isolating the individual merit of an attribute. Attribute selection is achieved by cross-validating the attribute rankings with respect to a classification learner to find the best attributes. Results are reported for a selection of standard data sets and two diverse learning schemes C4.5 and naive Bayes.

The evolution of plant functional variation: traits, spectra, and strategies

Abstract: Variation in plant functional traits results from evolutionary and environmental drivers that operate at a variety of different scales, which makes it a challenge to differentiate among them. In this article we describe patterns of functional trait variation and trait correlations within and among habitats in relation to several environmental and trade‐off axes. We then ask whether such patterns reflect natural selection and can be considered plant strategies. In so doing we highlight evidence that demonstrates that (1) patterns of trait variation across resource and environmental gradients (light, water, nutrients, and temperature) probably reflect adaptation, (2) plant trait variation typically involves multiple‐correlated traits that arise because of inevitable trade‐offs among traits and across levels of whole‐plant integration and that must be understood from a whole‐plant perspective, and (3) such adaptation may be globally generalizable for like conditions; i.e., the set of traits (collections of t...

MHC studies in nonmodel vertebrates: what have we learned about natural selection in 15 years?

Abstract: Elucidating how natural selection promotes local adaptation in interaction with migration, genetic drift and mutation is a central aim of evolutionary biology. While several conceptual and practical limitations are still restraining our ability to study these processes at the DNA level, genes of the major histocompatibility complex (MHC) offer several assets that make them unique candidates for this purpose. Yet, it is unclear what general conclusions can be drawn after 15 years of empirical research that documented MHC diversity in the wild. The general objective of this review is to complement earlier literature syntheses on this topic by focusing on MHC studies other than humans and mice. This review first revealed a strong taxonomic bias, whereby many more studies of MHC diversity in natural populations have dealt with mammals than all other vertebrate classes combined. Secondly, it confirmed that positive selection has a determinant role in shaping patterns of nucleotide diversity in MHC genes in all vertebrates studied. Yet, future tests of positive selection would greatly benefit from making better use of the increasing number of models potentially offering more statistical rigour and higher resolution in detecting the effect and form of selection. Thirdly, studies that compared patterns of MHC diversity within and among natural populations with neutral expectations have reported higher population differentiation at MHC than expected either under neutrality or simple models of balancing selection. Fourthly, several studies showed that MHC-dependent mate preference and kin recognition may provide selective factors maintaining polymorphism in wild outbred populations. However, they also showed that such reproductive mechanisms are complex and context-based. Fifthly, several studies provided evidence that MHC may significantly influence fitness, either by affecting reproductive success or progeny survival to pathogens infections. Overall, the evidence is compelling that the MHC currently represents the best system available in vertebrates to investigate how natural selection can promote local adaptation at the gene level despite the counteracting actions of migration and genetic drift. We conclude this review by proposing several directions where future research is needed.

On the Evolution of the Firm Size Distribution: Facts and Theory

Abstract: Using a comprehensive data set of Portuguese manufacturing firms, we show that the firm size distribution is significantly right-skewed, evolving over time toward a lognormal distribution. We also show that selection accounts for very little of this evolution. Instead, we propose a simple theory based on financing constraints. A calibrated version of our model does a good job at explaining the evolution of the firm size distribution. (JEL L11)

Toward an ecological synthesis: a case for habitat selection

Abstract: Habitat selection, and its associated density and frequency-dependent evolution, has a profound influence on such vital phenomena as population regulation, species interactions, the assembly of ecological communities, and the origin and maintenance of biodiversity. Different strategies of habitat selection, and their importance in ecology and evolution, can often be revealed simply by plots of density in adjacent habitats. For individual species, the strategies are closely intertwined with mechanisms of population regulation, and with the persistence of populations through time. For interacting species, strategies of habitat selection are not only responsible for species coexistence, but provide one of the most convenient mechanisms for measuring competition, and the various community structures caused by competitive interactions. Other kinds of interactions, such as those between predators and prey, demonstrate that an understanding of the coevolution of habitat-selection strategies among strongly interacting species is essential to properly interpret their spatial and temporal dynamics. At the evolutionary scale, the frequency dependence associated with habitat selection may often allow populations to diverge and diversify into separate species. Habitat selection thereby demonstrates how we can map microevolutionary strategies in behavior onto their population and community consequences, and from there, onto macroevolutionary patterns of speciation and adaptive radiation. We can anticipate that future studies of habitat selection will not only help us complete those maps, but that they will also continue to enrich the panoply of ideas that shape evolutionary ecology.

Frequentist Model Average Estimators

Abstract: The traditional use of model selection methods in practice is to proceed as if the final selected model had been chosen in advance, without acknowledging the additional uncertainty introduced by model selection. This often means underreporting of variability and too optimistic confidence intervals. We build a general large-sample likelihood apparatus in which limiting distributions and risk properties of estimators post-selection as well as of model average estimators are precisely described, also explicitly taking modeling bias into account. This allows a drastic reduction in complexity, as competing model averaging schemes may be developed, discussed, and compared inside a statistical prototype experiment where only a few crucial quantities matter. In particular, we offer a frequentist view on Bayesian model averaging methods and give a link to generalized ridge estimators. Our work also leads to new model selection criteria. The methods are illustrated with real data applications.

Effect of Recombination on the Accuracy of the Likelihood Method for Detecting Positive Selection at Amino Acid Sites

Abstract: Maximum-likelihood methods based on models of codon substitution accounting for heterogeneous selective pressures across sites have proved to be powerful in detecting positive selection in protein-coding DNA sequences. Those methods are phylogeny based and do not account for the effects of recombination. When recombination occurs, such as in population data, no unique tree topology can describe the evolutionary history of the whole sequence. This violation of assumptions raises serious concerns about the likelihood method for detecting positive selection. Here we use computer simulation to evaluate the reliability of the likelihood-ratio test (LRT) for positive selection in the presence of recombination. We examine three tests based on different models of variable selective pressures among sites. Sequences are simulated using a coalescent model with recombination and analyzed using codon-based likelihood models ignoring recombination. We find that the LRT is robust to low levels of recombination (with fewer than three recombination events in the history of a sample of 10 sequences). However, at higher levels of recombination, the type I error rate can be as high as 90%, especially when the null model in the LRT is unrealistic, and the test often mistakes recombination as evidence for positive selection. The test that compares the more realistic models M7 (beta) against M8 (beta and omega) is more robust to recombination, where the null model M7 allows the positive selection pressure to vary between 0 and 1 (and so does not account for positive selection), and the alternative model M8 allows an additional discrete class with omega = d(N)/d(S) that could be estimated to be >1 (and thus accounts for positive selection). Identification of sites under positive selection by the empirical Bayes method appears to be less affected than the LRT by recombination.

Overfitting in making comparisons between variable selection methods

Abstract: This paper addresses a common methodological flaw in the comparison of variable selection methods. A practical approach to guide the search or the selection process is to compute cross-validation performance estimates of the different variable subsets. Used with computationally intensive search algorithms, these estimates may overfit and yield biased predictions. Therefore, they cannot be used reliably to compare two selection methods, as is shown by the empirical results of this paper. Instead, like in other instances of the model selection problem, independent test sets should be used for determining the final performance. The claims made in the literature about the superiority of more exhaustive search algorithms over simpler ones are also revisited, and some of them infirmed.

An Estimator for Some Binary‐Outcome Selection Models Without Exclusion Restrictions

Abstract: This article provides a new maximum-likelihood estimator for selection models with dichotomous dependent variables when identical factors affect the selection equation and the equation of interest. Such situations arise naturally in game-theoretic models where selection is typically nonrandom and identical explanatory variables influence all decisions under investigation. When identical explanatory variables influence selection and a subsequent outcome of interest, the commonly used Heckman-type estimators identify from distributional assumptions about the residuals alone. When its own identifying assumption is reasonable, the new estimator allows the researcher to avoid the painful choice between identifying from distributional assumptions alone and adding a theoretically unjustified variable to the selection equation in a mistaken attempt to “boost” identification. The article uses Monte Carlo methods to compare the small-sample properties of the estimator with those of the Heckman-type estimator and ordinary probit.

Example selection for bootstrapping statistical parsers

Abstract: This paper investigates bootstrapping for statistical parsers to reduce their reliance on manually annotated training data. We consider both a mostly-unsupervised approach, cotraining, in which two parsers are iteratively re-trained on each other's output; and a semi-supervised approach, corrected co-training, in which a human corrects each parser's output before adding it to the training data. The selection of labeled training examples is an integral part of both frameworks. We propose several selection methods based on the criteria of minimizing errors in the data and maximizing training utility. We show that incorporating the utility criterion into the selection method results in better parsers for both frameworks.

Nonparametric Estimation of Sample Selection Models

Abstract: Sample selection models provide an important way of accounting for economic decisions that combine discrete and continuous choices and of correcting for nonrandom sampling. Nonparametric estimators for these models are developed in this paper. These can be used for estimating shapes and important economic quantities, as in standard nonparametnc regression. Endogeneity of regressors of interest is allowed for. Series estimators for these models are developed, which are useful for imposing additivity restnctions that arise from selection corrections. Convergence rates and asymptotic normality results are derived. An application to returns to schooling among Australian young females is given.

Inheritance and Natural Selection on Functional Traits

Abstract: We surveyed the literature published since 1985 for evidence of natural selection and heritability in vegetative functional traits and performance. Our goals were to (1) review patterns of selection on specific functional traits and (2) assess general evolutionary questions about selection and heritability for broad classes of traits. While generalizations about the functional significance of specific traits are premature, several functional hypotheses are supported. For example, herbivores can exert strong selection on secondary chemistry and mechanical defenses, but costs of resistance and negative correlations between defense traits may constrain their evolution. Competitive interactions select for early germination and favor stem elongation and shifts in flowering time where such responses actually minimize competitive effects. In the very few studies of physiology, selection on gas exchange and leaf size is clearly environment dependent. More generally, in reciprocal transplant experiments, populatio...

Perspective: detecting adaptive molecular polymorphism: lessons from the MHC.

Abstract: In the 1960s, when population geneticists first began to collect data on the amount of genetic variation in natural populations, balancing selection was invoked as a possible explanation for how such high levels of molecular variation are maintained. However, the predictions of the neutral theory of molecular evolution have since become the standard by which cases of balancing selection may be inferred. Here we review the evidence for balancing selection acting on the major histocompatibility complex (MHC) of vertebrates, a genetic system that defies many of the predictions of neutrality. We apply many widely used tests of neutrality to MHC data as a benchmark for assessing the power of these tests. These tests can be categorized as detecting selection in the current generation, over the history of populations, or over the histories of species. We find that selection is not detectable in MHC datasets in every generation, population, or every evolutionary lineage. This suggests either that selection on the MHC is heterogeneous or that many of the current neutrality tests lack sufficient power to detect the selection consistently. Additionally, we identify a potential inference problem associated with several tests of neutrality. We demonstrate that the signals of selection may be generated in a relatively short period of microevolutionary time, yet these signals may take exceptionally long periods of time to be erased in the absence of selection. This is especially true for the neutrality test based on the ratio of nonsynonymous to synonymous substitutions. Inference of the nature of the selection events that create such signals should be approached with caution. However, a combination of tests on different time scales may overcome such problems.

Gene selection: a Bayesian variable selection approach

Abstract: Selection of significant genes via expression patterns is an important problem in microarray experiments. Owing to small sample size and the large number of variables (genes), the selection process can be unstable. This paper proposes a hierarchical Bayesian model for gene (variable) selection. We employ latent variables to specialize the model to a regression setting and uses a Bayesian mixture prior to perform the variable selection. We control the size of the model by assigning a prior distribution over the dimension (number of significant genes) of the model. The posterior distributions of the parameters are not in explicit form and we need to use a combination of truncated sampling and Markov Chain Monte Carlo (MCMC) based computation techniques to simulate the parameters from the posteriors. The Bayesian model is flexible enough to identify significant genes as well as to perform future predictions. The method is applied to cancer classification via cDNA microarrays where the genes BRCA1 and BRCA2 are associated with a hereditary disposition to breast cancer, and the method is used to identify a set of significant genes. The method is also applied successfully to the leukemia data.

Interactive selection model for supplier selection process: An analytical hierarchy process approach

Abstract: Supplier Selection Process (SSP) becomes increasingly important for most manufacturing firms as it helps to reduce directly cost to the bottom line. The selection process involves the determination of quantitative and qualitative factors so as to select the best possible suppliers. It is essential to identify the relationship with the suppliers in terms of tangible factors. Owing to subjective human judgement in determining the relative importance of those selection factors, a method called Chain of Interaction is proposed to solve the problems associated with the dynamic nature of supply chain management. Focusing on the goodwill of an Analytic Hierarchy Process, an Interactive Selection Model is suggested to systemize the earlier steps, such as the determination of buyer-supplier relationships and formation of selection criteria, before the implementation of the Analytic Hierarchy Process with the help of Multi-Criterion Decision Making software called Expert Choice. The proposed Interactive Selection M...

Active Learning with support Vector machines in the drug discovery process

Abstract: We investigate the following data mining problem from computer-aided drug design: From a large collection of compounds, find those that bind to a target molecule in as few iterations of biochemical testing as possible. In each iteration a comparatively small batch of compounds is screened for binding activity toward this target. We employed the so-called “active learning paradigm” from Machine Learning for selecting the successive batches. Our main selection strategy is based on the maximum margin hyperplanegenerated by “Support Vector Machines”. This hyperplane separates the current set of active from the inactive compounds and has the largest possible distance from any labeled compound. We perform a thorough comparative study of various other selection strategies on data sets provided by DuPont Pharmaceuticals and show that the strategies based on the maximum margin hyperplane clearly outperform the simpler ones.

Using evolutionary algorithms as instance selection for data reduction in KDD: an experimental study

Abstract: Evolutionary algorithms are adaptive methods based on natural evolution that may be used for search and optimization As data reduction in knowledge discovery in databases (KDDs) can be viewed as a search problem, it could be solved using evolutionary algorithms (EAs) In this paper, we have carried out an empirical study of the performance of four representative EA models in which we have taken into account two different instance selection perspectives, the prototype selection and the training set selection for data reduction in KDD This paper includes a comparison between these algorithms and other nonevolutionary instance selection algorithms The results show that the evolutionary instance selection algorithms consistently outperform the nonevolutionary ones, the main advantages being: better instance reduction rates, higher classification accuracy, and models that are easier to interpret

Prices and Price Dispersion on the Web: Evidence from the Online Book Industry

Abstract: Using data collected between August, 1999, and January, 2000, covering 399 books, we examine pricing by thirty-two online United States-based bookstores. At the aggregate level, we find that both advertising and competitive structure had the predicted effects. More competition led to lower prices and to lower price dispersion. Holding competitive structure constant, more widely advertised items also had lower prices. At the firm level, we observe considerable heterogeneity in behavior. Firms had differentiated (or attempted to differentiate) on dimensions such as brand, price, and selection.

An integrated approach for supplier selection

Abstract: Competitive international business environment has forced many firms to focus on supply chain management to cope with highly increasing competition. Hence, supplier selection process has gained importance recently, since most of the firms have been spending considerable amount of their revenues on purchasing. The supplier selection problem involves conflicting multiple criteria that are tangible and intangible. Hence, the purpose of this study is to propose an integrated model for supplier selection. In order to achieve this purpose, supplier selection problem has been structured as an integrated lexicographic goal programming (LGP) and analytic hierarchy process (AHP) model including both quantitative and qualitative conflicting factors. The application process has been accomplished in a food company established in Istanbul, Turkey. In this study, the model building, solution and application processes of the proposed integrated model for supplier selection have been presented.

Measuring Nonlinear Selection

Abstract: A recent comprehensive review of empirical studies that measured the strength of selection concluded that there was little evidence for strong nonlinear selection in natural populations (Kingsolver et al. 2001). The median quadratic selection gradient identified by Kingsolver et al. (2001) was only 0.1, and gradients consistent with stabilizing or disruptive selection were found at a similar frequency and to be of similar magnitude. Stabilizing selection in particular is an important premise in many areas of evolutionary biology (Travis 1989), so this finding challenges our current understanding of how selection may operate in the wild (Conner 2001; Kingsolver et al. 2001). There is already some evidence that the strength and frequency of nonlinear selection identified in the review of Kingsolver et al. (2001) has influenced how evolutionary biologists view the potential role of stabilizing selection (Barton and Keightley 2002; Zhang et al. 2002). Kingsolver et al. (2001) suggest a number of reasons why empirical studies may not find or may underestimate nonlinear selection, including an empirical bias toward selecting experimental systems that are likely to show directional selection. In addition, nonlinear selection gradients are often able to be estimated from data sets but are not published (Kingsolver et al. 2001), probably as a consequence of the paucity of significant individual gradients and the difficulty of interpreting the overall pattern of nonlinear selection from the large number of estimated gradients. Here, we demonstrate that the strength of non-

Relevant document distribution estimation method for resource selection

Abstract: Prior research under a variety of conditions has shown the CORI algorithm to be one of the most effective resource selection algorithms, but the range of database sizes studied was not large. This paper shows that the CORI algorithm does not do well in environments with a mix of "small" and "very large" databases. A new resource selection algorithm is proposed that uses information about database sizes as well as database contents. We also show how to acquire database size estimates in uncooperative environments as an extension of the query-based sampling used to acquire resource descriptions. Experiments demonstrate that the database size estimates are more accurate for large databases than estimates produced by a competing method; the new resource ranking algorithm is always at least as effective as the CORI algorithm; and the new algorithm results in better document rankings than the CORI algorithm.

Natural selection and inheritance of breeding time and clutch size in the collared flycatcher

Abstract: Many characteristics of organisms in free-living populations appear to be under directional selection, possess additive genetic variance, and yet show no evolutionary response to selection. Avian breeding time and clutch size are often-cited examples of such characters. We report analyses of inheritance of, and selection on, these traits in a long-term study of a wild population of the collared flycatcher Ficedula albicollis. We used mixed model analysis with REML estimation (“animal models”) to make full use of the information in complex multigenerational pedigrees. Heritability of laying date, but not clutch size, was lower than that estimated previously using parent-offspring regressions, although for both traits there was evidence of substantial additive genetic variance (h2 = 0.19 and 0.29, respectively). Laying date and clutch size were negatively genetically correlated (rA = −0.41 ± 0.09), implying that selection on one of the traits would cause a correlated response in the other, but ther...

The Signature of Selection Mediated by Expression on Human Genes

Abstract: As the efficacy of natural selection is expected to be a function of population size, in humans it is usually presumed that selection is a weak force and hence that gene characteristics are mostly determined by stochastic forces. In contrast, in species with large population sizes, selection is expected to be a much more effective force. Evidence for this has come from examining how genic parameters vary with expression level, which appears to determine many of a gene's features, such as codon bias, amino acid composition, and size. However, not until now has it been possible to examine whether human genes show the signature of selection mediated by expression level. Here, then, to investigate this issue, we gathered expression data for >10,000 human genes from public data sets obtained by different technologies (SAGE and high-density oligonucleotide chip arrays) and compared them with gene parameters. We find that, even after controlling for regional effects, highly expressed genes code for smaller proteins, have less intronic DNA, and higher codon and amino acid biases. We conclude that, contrary to the usual supposition, human genes show signatures consistent with selection mediated by expression level.

Stability of the G-matrix in a population experiencing pleiotropic mutation, stabilizing selection, and genetic drift.

Abstract: Quantitative genetics theory provides a framework that predicts the effects of selection on a phenotype consisting of a suite of complex traits. However, the ability of existing theory to reconstruct the history of selection or to predict the future trajectory of evolution depends upon the evolutionary dynamics of the genetic variance-covariance matrix (G-matrix). Thus, the central focus of the emerging field of comparative quantitative genetics is the evolution of the G-matrix. Existing analytical theory reveals little about the dynamics of G, because the problem is too complex to be mathematically tractable. As a first step toward a predictive theory of G-matrix evolution, our goal was to use stochastic computer models to investigate factors that might contribute to the stability of G over evolutionary time. We were concerned with the relatively simple case of two quantitative traits in a population experiencing stabilizing selection, pleiotropic mutation, and random genetic drift. Our results show that G-matrix stability is enhanced by strong correlational selection and large effective population size. In addition, the nature of mutations at pleiotropic loci can dramatically influence stability of G. In particular, when a mutation at a single locus simultaneously changes the value of the two traits (due to pleiotropy) and these effects are correlated, mutation can generate extreme stability of G. Thus, the central message of our study is that the empirical question regarding G-matrix stability is not necessarily a general question of whether G is stable across various taxonomic levels. Rather, we should expect the G-matrix to be extremely stable for some suites of characters and unstable for others over similar spans of evolutionary time.

Genetic Variability at Neutral Markers, Quantitative Trait Loci and Trait in a Subdivided Population Under Selection

Abstract: Genetic variability in a subdivided population under stabilizing and diversifying selection was investigated at three levels: neutral markers, QTL coding for a trait, and the trait itself. A quantitative model with additive effects was used to link genotypes to phenotypes. No physical linkage was introduced. Using an analytical approach, we compared the diversity within deme (H(S)) and the differentiation (F(ST)) at the QTL with the genetic variance within deme (V(W)) and the differentiation (Q(ST)) for the trait. The difference between F(ST) and Q(ST) was shown to depend on the relative amounts of covariance between QTL within and between demes. Simulations were used to study the effect of selection intensity, variance of optima among demes, and migration rate for an allogamous and predominantly selfing species. Contrasting dynamics of the genetic variability at markers, QTL, and trait were observed as a function of the level of gene flow and diversifying selection. The highest discrepancy among the three levels occurred under highly diversifying selection and high gene flow. Furthermore, diversifying selection might cause substantial heterogeneity among QTL, only a few of them showing allelic differentiation, while the others behave as neutral markers.

Gaming device having a multiple selection and award distribution bonus scheme

Abstract: A gaming device having a multiple selection and award distribution bonus scheme. A selection is chosen from a group of selections. The game determines awards for distribution to the selection. Once determined, the awards are distributed to the selection and a player is provided with the awards. The game preferably utilizes a number of award pools in order to determine the award distribution. This award pool determination is based on, for example, a number of probability tables associated with the award pools.

Faking and selection: considering the use of personality from select-in and select-out perspectives.

Abstract: The effects of faking on criterion-related validity and the quality of selection decisions are examined in the present study by combining the control of an experiment with the realism of an applicant setting. Participants completed an achievement motivation measure in either a control group or an incentive group and then completed a performance task. With respect to validity, greater prediction error was found in the incentive condition among those with scores at the high end of the predictor distribution. When selection ratios were small, those in the incentive condition were more likely to be selected and had lower mean performance than those in the control group. Implications for using personality assessments from select-in and select-out strategies are discussed.

An identification algorithm for polynomial NARX models based on simulation error minimization

Abstract: Classical prediction error approaches for the identification of non-linear polynomial NARX/NARMAX models often yield unsatisfactory results for long-range prediction or simulation purposes, mainly due to incorrect or redundant model structure selection. The paper discusses some limitations of the standard approach and suggests two modifications: namely, a new index, based on the simulation error, is employed as the regressor selection criterion and a pruning mechanism is introduced in the model selection algorithm. The resulting algorithm is shown to be effective in the identification of compact and robust models, generally yielding model structures closer to the correct ones. Computational issues are also discussed. Finally, the identification algorithm is tested on a long-range prediction benchmark application.

Genetic similarity and quality interact in mate choice decisions by female mice

Abstract: Females express mate preferences for genetically dissimilar males, especially with respect to the major histocompatibility complex, MHC, and for males whose sexually selected signals indicate high genetic quality. The balance of selection pressure on each trait will depend on how females weight these desirable qualities under different conditions, but this has not been tested empirically. Here we show in mice that although MHC dissimilarity and a 'good genes' indicator (investment in scent-marking) both have a role in determining female preference, their relative influence can vary depending on the degree of variability in each trait among available males. Such interactions between condition-dependent and disassortative mate choice criteria suggest a mechanism by which female choice can contribute to maintenance of additive genetic variance in both the MHC and condition-dependent traits, even under consistent directional selection.