Algoritmos genéticos aplicados a um comitê de LS-SVM em problemas de classificação

The pattern classification is one of the machine learning subareas that has the most outstanding. Among the various approaches to solve pattern classification problems, the Support Vector Machines (SVM) receive great emphasis, due to its ease of use and good generalization performance. The Least Squares formulation of SVM (LS-SVM) finds the solution by solving a set of linear equations instead of quadratic programming implemented in SVM. The LS-SVMs provide some free parameters that have to be correctly chosen to achieve satisfactory results in a given task. Despite the LS-SVMs having high performance, lots of tools have been developed to improve them, mainly the development of new classifying methods and the employment of ensembles, in other words, a combination of several classifiers. In this work, our proposal is to use an ensemble and a Genetic Algorithm (GA), search algorithm based on the evolution of species, to enhance the LSSVM classification. In the construction of this ensemble, we use a random selection of attributes of the original problem, which it splits the original problem into smaller ones where each classifier will act. So, we apply a genetic algorithm to find effective values of the LS-SVM parameters and also to find a weight vector, measuring the importance of each machine in the final classification. Finally, the final classification is obtained by a linear combination of the decision values of the LS-SVMs with the weight vector. We used several classification problems, taken as benchmarks to evaluate the performance of the algorithm and compared the results with other classifiers

Estimação de Fasores para Proteção de Sistemas Elétricos Baseada em Mínimos Quadrados e Morfologia Matemática

This work proposes a new technique for phasor estimation applied in microprocessor numerical relays for distance protection of transmission lines, based on the recursive least squares method and called least squares modified random walking. The phasor estimation methods have compromised their performance, mainly due to the DC exponential decaying component present in fault currents. In order to reduce the influence of the DC component, a Morphological Filter (FM) was added to the method of least squares and previously applied to the process of phasor estimation. The presented method is implemented in MATLABr and its performance is compared to one-cycle Fourier technique and conventional phasor estimation, which was also based on least squares algorithm. The methods based on least squares technique used for comparison with the proposed method were: forgetting factor recursive, covariance resetting and random walking. The techniques performance analysis were carried out by means of signals synthetic and signals provided of simulations on the Alternative Transient Program (ATP). When compared to other phasor estimation methods, the proposed method showed satisfactory results, when it comes to the estimation speed, the steady state oscillation and the overshoot. Then, the presented method performance was analyzed by means of variations in the fault parameters (resistance, distance, angle of incidence and type of fault). Through this study, the results did not showed significant variations in method performance. Besides, the apparent impedance trajectory and estimated distance of the fault were analysed, and the presented method showed better results in comparison to one-cycle Fourier algorithm

Análise de séries temporais de coordenadas estimadas com GPS: uma proposta metodológica para detecção, remoção e recuperação de efeitos sazonais

GPS active networks are more and more used in geodetic surveying and scientific experiments, as water vapor monitoring in the atmosphere and lithosphere plate movement. Among the methods of GPS positioning, Precise Point Positioning (PPP) has provided very good results. A characteristic of PPP is related to the modeling and/or estimation of the errors involved in this method. The accuracy obtained for the coordinates can reach few millimeters. Seasonal effects can affect such accuracy if they are not consistent treated during the data processing. Coordinates time series analyses have been realized using Fourier or Harmonics spectral analyses, wavelets, least squares estimation among others. An approach is presented in this paper aiming to investigate the seasonal effects included in the stations coordinates time series. Experiments were carried out using data from stations Manaus (NAUS) and Fortaleza (BRFT) which belong to the Brazilian Continuous GPS Network (RBMC). The coordinates of these stations were estimated daily using PPP and were analyzed through wavelets for identification of the periods of the seasonal effects (annual and semi-annual) in each time series. These effects were removed by means of a filtering process applied in the series via the least squares adjustment (LSQ) of a periodic function. The results showed that the combination of these two mathematical tools, wavelets and LSQ, is an interesting and efficient technique for removal of seasonal effects in time series.

Orbit determination using GPS including perturbations due to geopotential coefficients of high degree and order and solar radiation pressure

The main target here is to determine the orbit of an artificial satellite, using signals of the GPS constellation and least squares algorithms implemented through sequential Givens rotations as a method of estimation, with the aim of improving the performance of the orbit estimation process and, at the same time, minimizing the computational procedure cost. Geopotential perturbations up to high order and direct solar radiation pressure were taken into account. It was also considered the position of the GPS antenna on the satellite body that, lately, consists of the influence of the satellite attitude motion in the orbit determination process. An application has been done, using real data from the Topex/Poseidon satellite, whose ephemeris is available at Internet. The best accuracy obtained in position was smaller than 5 meters for short period (2 hours) and smaller than 28 meters for long period (24 hours) orbit determination. In both cases, the perturbations mentioned before were taken into consideration and the analysis occurred without selective availability on the signals measurements.

Ferramentas quimiométricas aplicadas a classificação de amostras de leite em pó e quantificação de proteínas

In this work we used chemometric tools to classify and quantify the protein content in samples of milk powder. We applied the NIR diffuse reflectance spectroscopy combined with multivariate techniques. First, we carried out an exploratory method of samples by principal component analysis (PCA), then the classification of independent modeling of class analogy (SIMCA). Thus it became possible to classify the samples that were grouped by similarities in their composition. Finally, the techniques of partial least squares regression (PLS) and principal components regression (PCR) allowed the quantification of protein content in samples of milk powder, compared with the Kjeldahl reference method. A total of 53 samples of milk powder sold in the metropolitan areas of Natal, Salvador and Rio de Janeiro were acquired for analysis, in which after pre-treatment data, there were four models, which were employed for classification and quantification of samples. The methods employed after being assessed and validated showed good performance, good accuracy and reliability of the results, showing that the NIR technique can be a non invasive technique, since it produces no waste and saves time in analyzing the samples

Construção de modelos multivariados para determinação de lipídios totais e unidade em leite em pó comercial utilizando espectroscopia no infravermelho próximo

In this work calibration models were constructed to determine the content of total lipids and moisture in powdered milk samples. For this, used the near-infrared spectroscopy by diffuse reflectance, combined with multivariate calibration. Initially, the spectral data were submitted to correction of multiplicative light scattering (MSC) and Savitzsky-Golay smoothing. Then, the samples were divided into subgroups by application of hierarchical clustering analysis of the classes (HCA) and Ward Linkage criterion. Thus, it became possible to build regression models by partial least squares (PLS) that allowed the calibration and prediction of the content total lipid and moisture, based on the values obtained by the reference methods of Soxhlet and 105 ° C, respectively . Therefore, conclude that the NIR had a good performance for the quantification of samples of powdered milk, mainly by minimizing the analysis time, not destruction of the samples and not waste. Prediction models for determination of total lipids correlated (R) of 0.9955, RMSEP of 0.8952, therefore the average error between the Soxhlet and NIR was ± 0.70%, while the model prediction to content moisture correlated (R) of 0.9184, RMSEP, 0.3778 and error of ± 0.76%

Determinação do teor de diclofenaco sódico em comprimidos por espectroscopia no infravermelho próximo NIR com calibração multivariada PLS

This work is combined with the potential of the technique of near infrared spectroscopy - NIR and chemometrics order to determine the content of diclofenac tablets, without destruction of the sample, to which was used as the reference method, ultraviolet spectroscopy, which is one of the official methods. In the construction of multivariate calibration models has been studied several types of pre-processing of NIR spectral data, such as scatter correction, first derivative. The regression method used in the construction of calibration models is the PLS (partial least squares) using NIR spectroscopic data of a set of 90 tablets were divided into two sets (calibration and prediction). 54 were used in the calibration samples and the prediction was used 36, since the calibration method used was crossvalidation method (full cross-validation) that eliminates the need for a validation set. The evaluation of the models was done by observing the values of correlation coefficient R 2 and RMSEC mean square error (calibration error) and RMSEP (forecast error). As the forecast values estimated for the remaining 36 samples, which the results were consistent with the values obtained by UV spectroscopy

Espectroscopia no infravermelho próximo e métodos de calibração multivariada aplicados à determinação simultânea de parâmetros bioquímicos em plasma sanguíneo

In this work, the quantitative analysis of glucose, triglycerides and cholesterol (total and HDL) in both rat and human blood plasma was performed without any kind of pretreatment of samples, by using near infrared spectroscopy (NIR) combined with multivariate methods. For this purpose, different techniques and algorithms used to pre-process data, to select variables and to build multivariate regression models were compared between each other, such as partial least squares regression (PLS), non linear regression by artificial neural networks, interval partial least squares regression (iPLS), genetic algorithm (GA), successive projections algorithm (SPA), amongst others. Related to the determinations of rat blood plasma samples, the variables selection algorithms showed satisfactory results both for the correlation coefficients (R²) and for the values of root mean square error of prediction (RMSEP) for the three analytes, especially for triglycerides and cholesterol-HDL. The RMSEP values for glucose, triglycerides and cholesterol-HDL obtained through the best PLS model were 6.08, 16.07 e 2.03 mg dL-1, respectively. In the other case, for the determinations in human blood plasma, the predictions obtained by the PLS models provided unsatisfactory results with non linear tendency and presence of bias. Then, the ANN regression was applied as an alternative to PLS, considering its ability of modeling data from non linear systems. The root mean square error of monitoring (RMSEM) for glucose, triglycerides and total cholesterol, for the best ANN models, were 13.20, 10.31 e 12.35 mg dL-1, respectively. Statistical tests (F and t) suggest that NIR spectroscopy combined with multivariate regression methods (PLS and ANN) are capable to quantify the analytes (glucose, triglycerides and cholesterol) even when they are present in highly complex biological fluids, such as blood plasma

Determinação de parâmetros (sólidos solúveis, pH e acidez titulável) em ameixas intactas usando espectroscopia no infravermelho próximo e seleção de comprimento de onda

The aim of this study was to evaluate the potential of near-infrared reflectance spectroscopy (NIRS) as a rapid and non-destructive method to determine the soluble solid content (SSC), pH and titratable acidity of intact plums. Samples of plum with a total solids content ranging from 5.7 to 15%, pH from 2.72 to 3.84 and titratable acidity from 0.88 a 3.6% were collected from supermarkets in Natal-Brazil, and NIR spectra were acquired in the 714 2500 nm range. A comparison of several multivariate calibration techniques with respect to several pre-processing data and variable selection algorithms, such as interval Partial Least Squares (iPLS), genetic algorithm (GA), successive projections algorithm (SPA) and ordered predictors selection (OPS), was performed. Validation models for SSC, pH and titratable acidity had a coefficient of correlation (R) of 0.95 0.90 and 0.80, as well as a root mean square error of prediction (RMSEP) of 0.45ºBrix, 0.07 and 0.40%, respectively. From these results, it can be concluded that NIR spectroscopy can be used as a non-destructive alternative for measuring the SSC, pH and titratable acidity in plums

Uso da espectroscopia do infravermelho próximo e técnicas multivariadas para diferenciar escherichia coli e salmonella enteritidis inoculadas em polpa de fruta (abacaxi)

Aiming to consumer s safety the presence of pathogenic contaminants in foods must be monitored because they are responsible for foodborne outbreaks that depending on the level of contamination can ultimately cause the death of those who consume them. In industry is necessary that this identification be fast and profitable. This study shows the utility and application of near-infrared (NIR) transflectance spectroscopy as an alternative method for the identification and classification of Escherichia coli and Salmonella Enteritidis in commercial fruit pulp (pineapple). Principal Component Analysis (PCA), Independent Modeling of Class Analogy (SIMCA) and Discriminant Analysis Partial Least Squares (PLS-DA) were used in the analysis. It was not possible to obtain total separation between samples using PCA and SIMCA. The PLS-DA showed good performance in prediction capacity reaching 87.5% for E. coli and 88.3% for S. Enteritides, respectively. The best models were obtained for the PLS-DA with second derivative spectra treated with a sensitivity and specificity of 0.87 and 0.83, respectively. These results suggest that the NIR spectroscopy and PLS-DA can be used to discriminate and detect bacteria in the fruit pulp

Métodos estatísticos recursivos aplicados ao problema de estimação de vazão de gás em plantas de Plunger-lift

This work has as main objective to find mathematical models based on linear parametric estimation techniques applied to the problem of calculating the grow of gas in oil wells. In particular we focus on achieving grow models applied to the case of wells that produce by plunger-lift technique on oil rigs, in which case, there are high peaks in the grow values that hinder their direct measurement by instruments. For this, we have developed estimators based on recursive least squares and make an analysis of statistical measures such as autocorrelation, cross-correlation, variogram and the cumulative periodogram, which are calculated recursively as data are obtained in real time from the plant in operation; the values obtained for these measures tell us how accurate the used model is and how it can be changed to better fit the measured values. The models have been tested in a pilot plant which emulates the process gas production in oil wells

Effects of GHR gene polymorphisms on growth and carcass traits in Zebu and crossbred beef cattle

The growth hormone receptor (GHR) is the cell surface receptor for growth hormone (GH) and is required for GH to carry out its effects on target tissues. The objectives of the present study were to estimate the allele and genotype frequencies of the GHR/Alu I gene polymorphism located in the regulatory region in beef cattle belonging to different genetic groups and to determine associations between this polymorphism and growth and carcass traits. Genotyping was performed on 384 animals, including 79 Nellore (Zebu), 30 Canchim (5/8 Charolais+3/8 Zebu), 30 Simmental X Nellore crossbred and 245 Angus x Nellore crossbred cattle. Alleles Alu I(+), Alu I(-) and Alu I(N)-null allele-were evidenced for the GHR/Alu I polymorphism and the frequency of the Alu I(N) allele was significantly higher than the frequency of the Alu I(+) and Alu I(-) alleles in all genetic groups. Genotype Alu I(N/N) of the GHRIAlu I predominated in Nellore animals, while the Alu I(N/+) and Alu I(N/-) predominated in the other genetic groups. In the association studies, traits of interest were analyzed using the General Linear Model (GLM) procedure of the SAS program and least squares means of the genotypes were compared by the Tukey test. Significant associations (P < 0.05) were observed between the Alu I(N/N) genotype of the GHRIAlu I polymorphism and lower weight gain and body weight at slaughter, although a confounding between genotypes and genetic groups may have occurred. (c) 2005 Elsevier B.V. All rights reserved.

Parâmetros e tendência genética da produção de leite de cabra no Brasil

A caprinocultura leiteira no Brasil, apesar de ser uma atividade rural consolidada há algumas décadas, tem se mostrado totalmente dependente de outros países no que se refere ao melhoramento genético. A maioria dos plantéis existentes atualmente tem como base animais importados, e a renovação do material genético é feita por meio da importação de sêmen. Inexistem informações sobre o valor genético dos animais e sua evolução no decorrer dos anos. No presente trabalho, foram estimadas a herdabilidade e a repetibilidade da produção de leite utilizando o REML. Os valores obtidos foram 0,21557 e 0,21564, respectivamente. Para a predição do valor gênico dos animais, foi usado o procedimento BLUP com modelo animal. A mudança na tendência genética anual estimada por um modelo quadrático foi -0,8109 kg/ano², indicando desaceleração no ganho genético. A correlação de Pearson entre os valores gênicos dos bodes estimados com base na média da capacidade provável de produção das filhas obtida pelo método de mínimos quadrados com as estimadas pelas equações do modelo misto foi de 0,5751. A correlação de SPEARMAN entre as classificações dos bodes obtidos pelos dois métodos foi de 0,5813.

Fatores genéticos e de meio na produção de leite de caprinos leiteiros

Objetivou-se, no presente estudo, avaliar a produção de leite de caprinos leiteiros da região sudeste do Brasil, com intuito de verificar os fatores de meio e estimar os parâmetros genéticos pelo método dos mínimos quadrados (MMQ). Os controles de 1336 lactações foram inicialmente ajustados pela função multifásica (difásica) e calculou-se a produção de leite total (PLT). Os dados foram provenientes de sete propriedades e três raças (Parda Alpina, Saanen e Toggenburg). A média e o erro-padrão da PLT estimados pelo MMQ foram de 635,31 ±39,75 kg. A interação ano x estação do parto influenciou a PLT. em um dos anos estudados, a PLT foi menor para as cabras paridas no final da estação. Nas três estações de parto, observou-se comportamento quadrático da PLT, em função dos anos de parto. Para as três estações, a PLT aumentou de 1986 até meados de 1990, decrescendo em seguida. A idade de máxima PLT foi observada aos 46,65 meses. Das três raças estudadas, observou-se que as raças Parda Alpina e Saanen apresentaram alternância de superioridade na PLT em algumas fazendas, porém maiores que a Toggenburg. Os coeficientes de herdabilidade e repetibilidade da PLT estimados pelo MMQ foram de 0,296 ± 0,079 e 0,277 ± 0,033, respectivamente. Estes resultados revelam baixa confiabilidade em poucas observações dessas características ou na inconsistência das estimativas da função multifásica. Como a PLT é uma característica limitada ao sexo, sugere-se o teste de progênie como método de seleção mais eficiente para os reprodutores e uso de inseminação artificial como processo de disseminação do material genético selecionado.

Strategies to improve fertility in Bos indicus postpubertal heifers and nonlactating cows submitted to fixed-time artificial insemination

Two experiments were designed to evaluate strategies to increase fertility of Bos indicus postpubertal heifers and nonlactating cows submitted to a fixed-time artificial insemination (TAI) protocol consisting of an intravaginal device containing 1.9 g of progesterone (CIDR) insertion + estradiol benzoate on Day 0, CIDR withdrawal + estradiol cypionate on Day 9, and TAI on Day 11. In Experiment 1, heifers (n = 1153) received a new or an 18-d previously used CIDR and, on Day 9, prostaglandin F(2 alpha) (PGF(2 alpha)) + 0, 200, or 300 IU equine chorionic gonadotropin (eCG). Heifers treated with a new CIDR had greater (least squares means +/- SEM) serum concentration of progesterone on Day 9 (3.06 +/- 0.09 ng/mL vs. 2.53 +/- 0.09 ng/mL; P < 0.05) and a smaller follicle at TAI (11.61 +/- 0.11 nim vs. 12.05 +/- 0.12 mm; P < 0.05). Heifers with smaller follicles at TAI had lesser serum progesterone, concentrations on Day 18 and reduced rates of ovulation, conception, and pregnancy (P < 0.05). Treatment with eCG improved (P < 0.05) follicle diameter at TAI (11.50 +/- 0.10 mm, 11.90 +/- 0.11 mm, and 12.00 +/- 0.10 mm, for 0, 100, and 200 IU, respectively), serum progesterone concentration on Day 18 (2.77 +/- 0.11 ng/mL, 3.81 +/- 0.11 ng/mL, and 4.87 +/- 0.11 ng/mL), and rates of ovulation (83.8%, 88.5%, and 94.3%) and pregnancy (41.3%, 47.0%, and 46.7%). In Experiment 2, nonlactating Nelore cows (n = 702) received PGF(2 alpha) treatment on Days 7 or 9 and, on Day 9, 0 or 300 IU cCG. Cows receiving PGF(2 alpha) on Day 7 had lesser serum progesterone concentrations on Day 9 (3.05 +/- 0.21 ng/mL vs. 4.58 +/- 0.21 ng/mL; P < 0.05), a larger follicle at TAI (11.54 +/- 0.21 mm vs. 10.84 +/- 0.21 mm; P < 0.05), and improved (P < 0.05) rates of ovulation (85.4% vs. 77.0%), conception (60.9% vs. 47.2%), and pregnancy (52.0% vs. 36.4%). Treatment with eCG improved (P < 0.05) serum progesterone concentration on Day 18 (3.24 +/- 0.14 ng/mL vs. 4.55 +/- 0.14 ng/mL) and the rates of ovulation (72.4% vs. 90.0%) and pregnancy (37.5% vs. 50.8%). In conclusion, giving PGF(2 alpha) earlier in the protocol in nonlactating cows and eCG treatment in postpubertal heifers and nonlactating cows improved fertility in response to a TAI (progesterone + estradiol) protocol. (C) 2009 Elsevier B.V. All rights reserved.