Zugl.: Berlin, Techn. Univ., Diss., 2008

Att manuellt hantera och klassificera stora mängder textdokument tar mycket tid och kräver mycket personal, att göra detta med hjälp av maskininlärning är för ändamålet ett alternativ. Det här arbetet önskar ge läsaren en grundläggande inblick i hur automatisk klassificering av texter fungerar, samt ge en lätt samanställning av några av de vanligt förekommande algoritmerna för ändamålet. De exempel som visas använder sig av artiklar på engelska om teknik- och finansnyheter, men arbetet har avstamp i frågan om mognadsgrad av tekniken för hantering av svenska officiella dokument. Första delen är den vetenskapliga bakgrund som den andra delen vilar på, här beskrivs flera algoritmer och tekniker som sedan används i praktiska exempel. Rapporten ämnar inte beskriva en färdig produkt, utan fungerar så som ”proof of concept” för textklassificeringens användning. Avslutningsvis diskuteras resultaten från de tester som gjorts, och en av slutsatserna är att när det finns tillräckligt med data kan en enkel klassificerare prestera nästan likvärdigt med en tekniskt sett mer utvecklad och komplex klassificerare. Relateras prestandan hos klassificeraren till tidsåtgången visar detta på att komplexa klassificerare kräver hårdvara med hög beräkningskapacitet och mycket minne för att vara gångbara. / To manually handle and classify large quantities of text documents, takes a lot of time and demands a large staff, to use machine learning for this purpose is an alternative. This thesis aims to give the reader a fundamental insight in how automatic classification of texts work and give a quick overview of the most common algorithms used for this purpose. The examples that are shown uses news articles in English about tech and finance, but the thesis takes a start in the question about how mature the technique is for handling official Swedish documents. The first part is the scientific background on which the second part rests, here several algorithms and techniques are described which is used in practice later. The report does not aim to describe a product in any form but acts as a “proof of concept” for the use of text classification. Finally, the results from the tests are discussed, and one of the conclusions drawn is that when data is abundant a relatively simple classifier can perform close to equal to a technically more developed and complex classifier. If the performance of the classifier is related to the time taken this indicates that complex classifiers need hardware with high computational power and a fair bit of memory for the classifier to be viable.

This dissertation focuses on two distinct but related problems involving detection in multiple images. The first problem focuses on the accurate detection of runways by fusing Synthetic Vision System (SVS) and Enhanced Vision System (EVS) images. A novel procedure is developed to accurately detect runways and horizons and also enhance runway surrounding areas by fusing enhanced vision system (EVS) and synthetic vision system (SVS) images of the runway while an aircraft is landing. Because the EVS and SVS frames are not aligned, a registration step is introduced to align the EVS and SVS images prior to fusion. The most notable feature of the registration procedure is that it is guided by the information extracted from the weather-invariant SVS images. Four fusion rules based on combining Discrete Wavelet Transform (DWT) sub-bands are implemented and evaluated. The resulting procedure is tested on real EVS-SVS image pairs and also on image pairs containing simulated EVS images with varying levels of turbulence. The subjective and objective evaluations reveal that runways and horizons can be detected accurately even in poor visibility conditions. Furthermore, it is demonstrated that different aspects of the EVS and SVS images can be emphasized by using different DWT fusion rules. Another notable feature is that the entire procedure is autonomous throughout the landing sequence irrespective of the weather conditions. Given the excellent fusion results and the autonomous feature, it can be concluded that the fusion procedure developed is quite promising for incorporation into head-up displays (HUDs) to assist pilots in safely landing aircrafts in varying weather conditions. The second problem focuses on the blind detection of hidden messages that are embedded in images using various steganography methods. A new steganalysis strategy is introduced to blindly detect hidden messages that have been embedded in JPEG images using various steganography techniques. The key contribution is the formulation of a multi-domain feature extraction, ranking, and selection strategy to improve the steganalysis performance. The multi-domain features are statistical measures extracted from DWT, muti-wavelet (MWT), and slantlet (SLT) transforms. Feature ranking and selection is based on evaluating the performance of each feature independently and combining the best uncorrelated features. The resulting feature set is used in conjunction with discriminant analysis and support vector classifiers to detect the presence/absence of hidden messages in images. Numerous experiments are conducted to demonstrate the improved performance of the new steganalysis strategy over existing steganalysis methods.

Traditional forecasting approaches forecast the total system load directly without considering the individual consumer's load. With the introduction of the smart grid, lots of renewable energy resources such as wind and solar are added to the system from consumer side fluctuates the system load and makes forecasting more complex. Thus, it is necessary to forecast individual consumers load. Here, a framework is presented in which individual customer loads is forecasted rather than the system load. At first, a hierarchical cluster analysis is performed to classify daily load patterns into different groups for all the individuals. Then an association analysis is performed to determine critical influential factors that affect the load curve for given day. The next step is the application of a decision tree to establish classification rules between the different groups of the load curve and the critical influential factors. Then, appropriate forecasting models are chosen for different load patterns and the individual load is forecasted. Finally, the forecasted total system load is obtained through an aggregation of an individual load forecasting results. The relative error of forecasting the system load using this framework is compared with the relative errors using SVM regression and this framework had better accuracy. This framework is also used for forecasting the power output of the renewable generation. Also, the results of the day ahead forecast of system load and renewable generation is used for economic power scheduling for the microgrid and peak shaving for the utilities.

As diversas áreas da Engenharia, em parceria com a ciência médica, têm contribuído de forma eficaz para o avanço do conhecimento e dos resultados em aplicações práticas na vida do deficiente físico. De forma geral, pesquisas com este foco têm permitido o desenvolvimento de dispositivos e recursos com o objetivo de oferecer novamente a mobilidade e a liberdade perdidas com a deficiência. Este trabalho tem a finalidade de desenvolver um sistema que utiliza Eletromiografia de Superfície e Máquina de Vetores de Suporte para a caracterização de determinados movimentos de um braço humano, possibilitando, futuramente, a integração em sistemas de reabilitação. Primeiramente os sinais mioelétricos são obtidos nos músculos do braço de voluntários através de eletrodos de superfície ligados a um eletromiógrafo. O sinal é adquirido, utilizando como padrão um modelo virtual que demonstra ao voluntário os movimentos do segmento mão-braço que devem ser imitados. Esses movimentos são executados e seus sinais mioelétricos adquiridos. Posteriormente, esses sinais são processados e características são extraídas. Em seguida, são alocadas algumas de suas características (RMS, média, variância, desvio padrão, skewness e kurtosis) na entrada da Máquina de Vetores de Suporte, que apresenta, como saída, o reconhecimento, ou não, do movimento previamente executado pelo voluntário. No final do processo, observou-se que aumentando o número de canais elevou-se a taxa de acerto dos movimentos e, com a retirada de determinada característica, houve decréscimo na taxa de acerto do sistema. Nestes casos, os 9 movimentos distintos atingiram uma taxa de acerto média de 83,2%, para dois canais, e 91,3%, para oito canais, e, em ambos sistemas de canais, com as seis características. / A wide range of engineering scopes, along with the knowledge from the medical science, has efficiently been contributing to further knowledge and results for practical applications in the life of the physically challenged. In general, such researches have allowed the development of devices and resources aimed at giving back the mobility and freedom lost with the deficiency. This paper intends to develop a system that uses Surface Electromyography and Support-Vector Machines (SVM) for the characterization of specific movements of a human arm enabling the future integration in rehabilitation systems. At first, myoelectric signals are obtained in the arm muscles of volunteers by means of surface electrodes attached to an Electromyography. The signal is acquired using a virtual model as pattern demonstrating to the volunteer the hand-arm movements which are to be replicated by the subject. As these movements are done, its respective myoelectric signals are acquired. Later on, these signals are processed and their characteristics extracted. Some of these features (such as RMS, standard deviation, variance, mean, kurtosis, skewness) will then be inserted in as input data in the Support- Vector Machine, which shows as an output a valid or null recognition of the movement earlier executed by the volunteer. At the end of the process, it was observed that increasing the number of channels increased by hit rate movements and, with the removal of certain characteristic, there was a decrease in the hit rate of the system. In these cases, nine distinct movements reached an average accuracy of 83.2% for two channels, and 91.3% for eight channels, and in both systems of channels, with six features.

Submitted by CÍCERO MANOEL DOS SANTOS null (ciceromanoel2007@gmail.com) on 2016-04-05T16:44:35Z No. of bitstreams: 1 Tese_FINAL.pdf: 8502485 bytes, checksum: 26494d2583b4a89dbd92a2f0a37703b8 (MD5) / Approved for entry into archive by Juliano Benedito Ferreira (julianoferreira@reitoria.unesp.br) on 2016-04-07T19:37:27Z (GMT) No. of bitstreams: 1 santos_cm_dr_bot.pdf: 8502485 bytes, checksum: 26494d2583b4a89dbd92a2f0a37703b8 (MD5) / Made available in DSpace on 2016-04-07T19:37:27Z (GMT). No. of bitstreams: 1 santos_cm_dr_bot.pdf: 8502485 bytes, checksum: 26494d2583b4a89dbd92a2f0a37703b8 (MD5) Previous issue date: 2016-03-04 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / A irradiação direta na incidência normal (Hb) possui papel importante no manejo de culturas agrícolas, na utilização como fonte de energia renovável e na modelagem atmosférica. Apesar de sua importância em diferentes áreas, medidas pontuais de Hb não são facilmente disponíveis nos centros de pesquisas, devido ao elevado custo de exportação dos sensores e suas manutenções periódicas. Os modelos estatísticos têm sido desenvolvidos e utilizados para estimativa de Hb nos locais onde não são monitorados. Estes modelos, normalmente, utilizam a Hg como variável de entrada, pois é a variável mais comumente medida em estações solarimétricas. Os modelos estatísticos correlacionam à fração transmitida da irradiação direta na incidência normal (ktb) com transmissividade atmosférica (kt) ou com a razão de insolação (n/N). Recentemente as técnicas de Aprendizado de Máquinas foram inseridas para estimativa de Hb. Teoricamente, são técnicas que apresentam alto desempenho na estimativa de modelos e gerar valores estimados mais precisos de Hb que os modelos estatísticos. O trabalho está divido em 4 capítulos divididos da seguinte forma. Capítulo 1: Propor a utilização da técnica Máquina de Vetor de Suporte – SVM e da Redes Neurais Artificiais para estimativa de Hb e comparar com os modelos estatísticos, testando diferentes variáveis de entrada, . Capítulo 2: Comparar a SVM com os modelos estatísticos. Capítulo 3: Comparar Rede Neural Artificial – RNA com os modelos estatísticos, utilizando o algoritmo Backpropagation. Capítulo 4: Modelagem da turbidez atmosférica de Linke com Hb. A fração transmitida de Hb (ktb) é modelada para obter Hb. Para treinamento e validação dos modelos é utilizado um banco de dados de 13 anos (1996 – 2008), medidos na estação radiométrica localizada na Faculdade de Ciências Agronômicas – FCA/UNESP (22,85°S; 48,45°W e 786m). Foram testadas diferentes variáveis de entrada para verificar qual a melhor na estimava dos modelos. Os índices estatísticos: MBE, rMBE, RMSE, rRMSE, d de Willmott e o erro percentual (%) são utilizados para validar os modelos. Os modelos foram propostos e avaliados nas partições de tempo: horária e diária. Os resultados mostraram que os modelos estatísticos estimam Hb com resultados (20% ≤ rRMSE < 30%). Os modelos propostos (SVM e RNA) geram resultados melhores que os modelos estatísticos e são indicados para estimativa de Hb (rRMSE < 20%). O modelo da SVM estima Hb melhor que RNA, por isso seu uso é tido como a primeira escolha entre os modelos. / The direct irradiation at normal incidence (Hb) is an important role in the management of crops, in the use as a renewable energy source and atmospheric modeling. Despite its importance in different areas, specific measures Hb are not readily available in research centers, due to the high cost of exporting the sensors and periodic maintenance of the sensors. Statistical models have been developed and used to estimate Hb in places where they are not monitored. These models usually use the Hg as input variable, as is the variable most commonly measured in solarimetric stations. Statistical models correlate to the fraction transmitted at Hb (ktb) with atmospheric transmissivity (kt) or insolation ratio (n/N). Recently the Machine Learning techniques (ML) were inserted for estimation of Hb. Theoretically, these techniques have greater capacity to model and generate more precise values of Hb that statistical models. The work is divided into four chapters divided as follows. Chapter 1: To propose the use of Support Vector Machine (SVM) and Artificial Neural Networks (ANN) technical to estimate Hb and compare the statistical models, testing different input variables. Chapter 2: To compare the SVM with the statistical models. Chapter 3: To compare Artificial Neural Network ANN) with statistical models using the backpropagation algorithm. Chapter 4: Modeling of atmospheric turbidity Linke with Hb. The ktb is modeled for get indirectly Hb. The validation methodology of the models with typical and atypical year is adopted and evaluated. It used a database of 13 years data (1996-2008), measured in radiometric station located at the Faculty of Agricultural Sciences - FCA/UNESP (22.85° S, 48.45° W and 786m. Different input variables are tested in the models to see if the estimate is improving. The variables used are: Hb, Hg, solar insolation (n), air temperature and relative humidity the other variables were obtained by mathematical equations. Statistical indices: MBE, rMBE, RMSE, rRMSE, d Willmontt and percent error (%) are used to validate the models. The models are proposed and evaluated in time: hourly and daily partitions. The results show that the statistical models estimate Hb with acceptable results (rRMSE ≤ 20% <30%). The proposed models (SVM and ANN) generate better results than the statistical models and are suitable for estimation of Hb (rRMSE <20%). The model of SVM estimates Hb better than ANN, so its use is considered the first choice among the models. / CNPq: 140104/2013-5

abstract: Robust and stable decoding of neural signals is imperative for implementing a useful neuroprosthesis capable of carrying out dexterous tasks. A nonhuman primate (NHP) was trained to perform combined flexions of the thumb, index and middle fingers in addition to individual flexions and extensions of the same digits. An array of microelectrodes was implanted in the hand area of the motor cortex of the NHP and used to record action potentials during finger movements. A Support Vector Machine (SVM) was used to classify which finger movement the NHP was making based upon action potential firing rates. The effect of four feature selection techniques, Wilcoxon signed-rank test, Relative Importance, Principal Component Analysis, and Mutual Information Maximization was compared based on SVM classification performance. SVM classification was used to examine the functional parameters of (i) efficacy (ii) endurance to simulated failure and (iii) longevity of classification. The effect of using isolated-neuron and multi-unit firing rates was compared as the feature vector supplied to the SVM. The best classification performance was on post-implantation day 36, when using multi-unit firing rates the worst classification accuracy resulted from features selected with Wilcoxon signed-rank test (51.12 ± 0.65%) and the best classification accuracy resulted from Mutual Information Maximization (93.74 ± 0.32%). On this day when using single-unit firing rates, the classification accuracy from the Wilcoxon signed-rank test was 88.85 ± 0.61 % and Mutual Information Maximization was 95.60 ± 0.52% (degrees of freedom =10, level of chance =10%) / Dissertation/Thesis / Masters Thesis Bioengineering 2015

O diagnóstico de doenças neuromusculares é feito pelo uso conjunto de várias ferramentas. Dentre elas, o exame de eletromiografia clínica fornece informações vitais ao diagnóstico. A aplicação de alguns classificadores (discriminante linear e redes neurais artificiais) aos diversos parâmetros dos sinais de eletromiografia (número de fases, de reversões e de cruzamentos de zero, freqüência mediana, coeficientes auto-regressivos) tem fornecido resultados promissores na literatura. No entanto, a necessidade de um número grande de coeficientes auto-regressivos direcionou este mestrado ao uso de modelos auto-regressivos de média móvel com um número menor de coeficientes. A classificação (em normal, neuropático ou miopático) foi feita pela máquina de vetor de suporte, um tipo de rede neural artificial de uso recente. O objetivo deste trabalho foi o de estudar a viabilidade do uso de modelos auto-regressivos de média móvel (ARMA) de ordem baixa, em vez de auto-regressivos de ordem alta, em conjunção com a máquina de vetor de suporte, para auxílio ao diagnóstico. Os resultados indicam que a máquina de vetor de suporte tem desempenho melhor que o discriminante linear de Fisher e que os modelos ARMA(1,11) e ARMA(1,12) fornecem altas taxas de classificação (81,5%), cujos valores são próximos ao máximo obtido com modelos auto-regressivos de ordem 39. Portanto, recomenda-se o uso da máquina de vetor de suporte e de modelos ARMA (1,11) ou ARMA(1,12) para a classificação de sinais de eletromiografia de agulha, de 800ms de duração e amostrados a 25kHz. / The diagnosis of neuromuscular diseases is attained by the combined use of several tools. Among these tools, clinical electromyography provides key information to the diagnosis. In the literature, the application of some classifiers (linear discriminant and artificial neural networks) to a variety of electromyography parameters (number of phases, turns and zero crossings; median frequency, auto-regressive coefficients) has provided promising results. Nevertheless, the need of a large number of auto-regressive coefficients has guided this Master\'s thesis to the use of a smaller number of auto-regressive moving-average coefficients. The classification task (into normal, neuropathic or myopathic) was achieved by support vector machines, a type of artificial neural network recently proposed. This work\'s objective was to study if low-order auto-regressive moving-average (ARMA) models can or cannot be used to substitute high-order auto-regressive models, in combination with support vector machines, for diagnostic purposes. Results point that support vector machines have better performance than Fisher linear discriminants. They also show that ARMA(1,11) and ARMA(1,12) models provide high classification rates (81.5%). These values are close to the maximum obtained by using 39 auto-regressive coefficients. So, we recommend the use of support vector machines and ARMA(1,11) or ARMA(1,12) to the classification of 800ms needle electromyography signals acquired at 25kHz.

Submitted by BARTOLOMEU FÉLIX TANGUNE null (tanguneb@gmail.com) on 2017-05-31T13:12:46Z No. of bitstreams: 1 Bartolomeu Felix Tangune_tese.pdf: 3390592 bytes, checksum: 0daf84bae7e268e5ff6b06e039ea9043 (MD5) / Approved for entry into archive by Luiz Galeffi (luizgaleffi@gmail.com) on 2017-05-31T18:38:01Z (GMT) No. of bitstreams: 1 tangune_bf_dr_bot.pdf: 3390592 bytes, checksum: 0daf84bae7e268e5ff6b06e039ea9043 (MD5) / Made available in DSpace on 2017-05-31T18:38:01Z (GMT). No. of bitstreams: 1 tangune_bf_dr_bot.pdf: 3390592 bytes, checksum: 0daf84bae7e268e5ff6b06e039ea9043 (MD5) Previous issue date: 2017-05-08 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / A evapotranspiração de referência (ETo) é importante na agricultura para satisfazer as necessidades de água das culturas e para o manejo dos sistemas de irrigação. A ETo pode ser estimada com precisão a partir do método padrão de Penman Monteith FAO 56, porém, o seu uso é bastante complexo. Sendo assim, vários métodos empíricos de uso simples vem sendo desenvolvidos por diversos pesquisadores, todavia, a sua escolha deve ser feita de forma cuidadosa, pois apresentam um desempenho que varia em função das condições climáticas de cada local. A variabilidade do desempenho dos métodos empíricos tem levado os pesquisadores a procurarem outros métodos alternativos. Como resultado dessas pesquisas, há que destacar a técnica de aprendizado de máquinas (TAM): redes neurais artificiais (RNAs) e máquina vetor de suporte (MVS). Diante do exposto, o presente trabalho foi dividido em três capítulos, onde no primeiro capítulo foi avaliado o desempenho dos métodos empíricos de temperatura (Benevides e Lopez - BenL, Hamon -Ham, Blaney Criddle Original e Hargreaves Samani -HS) e de radiação solar (Abtew, Jensen Haise - JensH, Makkink e Irmak) na estimativa da ETo no estado de São Paulo. Todos os métodos foram avaliados em relação ao método padrão em escala anual e sazonal. Os resultados obtidos na escala anual mostraram que o método de Abtew apresentou o melhor desempenho. Na escala sazonal, observou-se que o método de JensH foi melhor no inverno, o de Irmak e de Abtew no verão e outono. O método de Abtew foi também melhor na primavera. No segundo capítulo, foi avaliado o desempenho dos métodos de HS, e de Abtew (melhores métodos empíricos em escala anual), RNAs e MVS. A RNA utilizada foi do tipo Multilayer Perceptron, com algoritmo de aprendizado Backpropagation e na MVS utilizou-se a função Radial Basic Function de Kernel, com algoritmo Regression Sequential Minimal Optimization. Os resultados obtidos na escala anual mostraram que a R6 (da RNA) e a M6 (da MVS) compostas por temperatura máxima (Tmax), mínima (Tmin), média do ar (T), radiação extraterrestre (Ra) e Rs produziram o melhor desempenho. Na escala sazonal, o melhores resultados foram observados nas arquiteturas R3 e M3, R4 e M4, R5 e M5, R6 e M6, compostas por: Tmax, Tmin, T, Ra e velocidade do vento; Tmax, Tmin, T, Ra e umidade relativa do ar; T e Rs, respectivamente. Tanto no capítulo 1 quanto no 2, as análises estatísticas foram feitas com base nos índices MBE (Mean Bias Error), RSME (Root Mean Square Error), “d” de Willmott e R2 (coeficiente de determinação). No terceiro capítulo, foi avaliada a técnica de interpolação por krigagem ordinária pontual (KOP), cujos variogramas obtidos foram avaliados com base na soma dos quadrados dos resíduos, em escala anual e sazonal. Todos os modelos variográficos obtidos apresentaram uma dependência espacial forte. A posterior, fez-se a validação cruzada da KOP com base nos coeficientes angular e linear da reta de regressão linear simples, MBE, RSME e MSDR (Mean squared deviation ratio ), cujos resultados mostraram um ótimo desempenho da KOP. / The reference evapotranspiration (ETo) is important in agriculture for crop water management and irrigation systems management. The ETo can be estimated accurately by the FAO 56 standard method of Penman Monteith, however, its use is complex. Thus, several empirical methods of simple use have been developed by many researchers, but their choice must be made carefully because they present a performance that change according to the climate conditions of each location. The variability of the performance of empirical methods has led researchers to look for alternative methods. As the result, we must highlight the machine learning technique (MLT), such as artificial neural networks (ANNs) and support vector machine (SVM). This work was divided into three chapters. In the first chapter, four temperature- based (Benevides e Lopez - BenL, Hamon -Ham, Blaney Criddle Original e Hargreaves Samani -HS) and four radiation- based (Abtew, Jensen Haise - JensH, Makkink and Irmak) ETo methods were tested against FAO 56 method, using annual and seasonal scale in the state of São Paulo. The results obtained in the annual scale showed that the Abtew method presented the best performance. On the seasonal scale, it was observed that the JensH method was better in the winter, the Irmak and Abtew methods were better in the summer and autumn. The Abtew method was also better in the spring. In the second chapter, HS and Abtew methods, ANNs and SVM were used. The ANN used was Multilayer Perceptron with Backpropagation learning algorithm, and in the SVM, was used Kernel Radial Basic Function with Regression Sequential Minimal Optimization learning algorithm. The obtained results in the annual scale showed that R6 for RNA and M6 for MVS composed of maximum temperature (Tmax), minimum temperature (Tmin), average air temperature (T), extraterrestrial radiation (Ra) and global solar radiation (Rs) had a better performance. On the seasonal scale, the better performance was observed in R3 e M3, R4 e M4, R5 e M5, R6 e M6 architectures, composed of Tmax, Tmin, T, Ra and wind speed; Tmax, Tmin, T, Ra and relative humidity); T and Rs; R6 and M6, respectively. All methods were analyzed using MBE (Mean Bias Error), RMSE (Root Mean Square Error), “d” of Wilmot (1985) and R2 (determination coefficient). In the third chapter, the technique of interpolation by ordinary punctual kriging (OPK) was evaluated, whose variograms were evaluated based on the residuals sum of squares, on an annual and seasonal scale. All the variographic models obtained showed a strong spatial dependence. Afterwards, cross-validation of OPK was performed based on the angular (β1) and linear (βo) coefficients of the simple linear regression line, MBE, RSME and MSDR (Mean squared deviation ratio), whose results showed an excellent performance of OPK.

Made available in DSpace on 2014-12-17T14:55:12Z (GMT). No. of bitstreams: 1 AdrianaT_TESE.pdf: 618602 bytes, checksum: 8ea994949daea03408599ce92036681a (MD5) Previous issue date: 2012-09-26 / The Support Vector Machines (SVM) has attracted increasing attention in machine learning area, particularly on classification and patterns recognition. However, in some cases it is not easy to determinate accurately the class which given pattern belongs. This thesis involves the construction of a intervalar pattern classifier using SVM in association with intervalar theory, in order to model the separation of a pattern set between distinct classes with precision, aiming to obtain an optimized separation capable to treat imprecisions contained in the initial data and generated during the computational processing. The SVM is a linear machine. In order to allow it to solve real-world problems (usually nonlinear problems), it is necessary to treat the pattern set, know as input set, transforming from nonlinear nature to linear problem. The kernel machines are responsible to do this mapping. To create the intervalar extension of SVM, both for linear and nonlinear problems, it was necessary define intervalar kernel and the Mercer s theorem (which caracterize a kernel function) to intervalar function / As m?quinas de vetores suporte (SVM - Support Vector Machines) t?m atra?do muita aten??o na ?rea de aprendizagem de m?quinas, em especial em classifica??o e reconhecimento de padr?es, por?m, em alguns casos nem sempre ? f?cil classificar com precis?o determinados padr?es entre classes distintas. Este trabalho envolve a constru??o de um classificador de padr?es intervalar, utilizando a SVM associada com a teoria intervalar, de modo a modelar com uma precis?o controlada a separa??o entre classes distintas de um conjunto de padr?es, com o objetivo de obter uma separa??o otimizada tratando de imprecis?es contidas nas informa??es do conjunto de padr?es, sejam nos dados iniciais ou erros computacionais. A SVM ? uma m?quina linear, e para que ela possa resolver problemas do mundo real, geralmente problemas n?o lineares, ? necess?rio tratar o conjunto de padr?es, mais conhecido como conjunto de entrada, de natureza n?o linear para um problema linear, as m?quinas kernels s?o respons?veis por esse mapeamento. Para a extens?o intervalar da SVM, tanto para problemas lineares quanto n?o lineares, este trabalho introduz a defini??o de kernel intervalar, bem como estabelece o teorema que valida uma fun??o ser um kernel, o teorema de Mercer para fun??es intervalares