Orientador: Prof. Dr. Ricardo Suyama / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, Santo André, 2018. / De forma geral a Classificação de Padrões voltada a Processamento de Sinais vem sendo estudada e utilizada para a interpretação de informações diversas, que se manifestam em forma de imagens, áudios, dados geofísicos, impulsos elétricos, entre outros. Neste trabalho são estudadas técnicas de Machine Learning aplicadas ao problema de identificação de instrumentos musicais, buscando obter um sistema automático de reconhecimento de timbres. Essas técnicas foram utilizadas especificamente com cinco instrumentos da categoria de Sopro de Madeira (o Clarinete, o Fagote, a Flauta, o Oboé e o Sax). As técnicas utilizadas foram o kNN (com k = 3) e o SVM (numa configuração não linear), assim como foram estudadas algumas características (features) dos áudios, tais como o MFCC (do inglês Mel-Frequency Cepstral Coefficients), o ZCR (do inglês Zero Crossing Rate), a entropia, entre outros, sendo fonte de dados para os processos de treinamento e de teste. Procurou-se estudar instrumentos nos quais se observa uma aproximação nos timbres, e com isso verificar como é o comportamento de um sistema classificador nessas condições específicas. Observou-se também o comportamento dessas técnicas com áudios desconhecidos do treinamento, assim como com trechos em que há uma mistura de elementos (gerando interferências para cada modelo classificador) que poderiam desviar os resultados, ou com misturas de elementos que fazem parte das classes observadas, e que se somam num mesmo áudio. Os resultados indicam que as características selecionadas possuem informações relevantes a respeito do timbre de cada um dos instrumentos avaliados (como observou-se em relação aos solos), embora a acurácia obtida para alguns dos instrumentos tenha sido abaixo do esperado (como observou-se em relação aos duetos). / In general, Pattern Classification for Signal Processing has been studied and used for the interpretation of several information, which are manifested in many ways, like: images, audios, geophysical data, electrical impulses, among others. In this project we study techniques of Machine Learning applied to the problem of identification of musical instruments, aiming to obtain an automatic system of timbres recognition. These techniques were used specifically with five instruments of Woodwind category (Clarinet, Bassoon, Flute, Oboe and Sax). The techniques used were the kNN (with k = 3) and the SVM (in a non-linear configuration), as well as some audio features, such as MFCC (Mel-Frequency Cepstral Coefficients), ZCR (Zero Crossing Rate), entropy, among others, used as data source for the training and testing processes. We tried to study instruments in which an approximation in the timbres is observed, and to verify in this case how is the behavior of a classifier system in these specific conditions. It was also observed the behavior of these techniques with audios unknown to the training, as well as with sections in which there is a mixture of elements (generating interferences for each classifier model) that could deviate the results, or with mixtures of elements that are part of the observed classes, and added in a same audio. The results indicate that the selected characteristics have relevant information regarding the timbre of each one of evaluated instruments (as observed on the solos results), although the accuracy obtained for some of the instruments was lower than expected (as observed on the duets results).

Submitted by Geyciane Santos (geyciane_thamires@hotmail.com) on 2015-07-15T14:33:14Z No. of bitstreams: 1 Dissertação - Adriano Mendes Gil.pdf: 11255112 bytes, checksum: 36763272079e769a8fc63a58ab9d3461 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-07-20T13:59:00Z (GMT) No. of bitstreams: 1 Dissertação - Adriano Mendes Gil.pdf: 11255112 bytes, checksum: 36763272079e769a8fc63a58ab9d3461 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-07-20T14:04:10Z (GMT) No. of bitstreams: 1 Dissertação - Adriano Mendes Gil.pdf: 11255112 bytes, checksum: 36763272079e769a8fc63a58ab9d3461 (MD5) / Made available in DSpace on 2015-07-20T14:04:10Z (GMT). No. of bitstreams: 1 Dissertação - Adriano Mendes Gil.pdf: 11255112 bytes, checksum: 36763272079e769a8fc63a58ab9d3461 (MD5) Previous issue date: 2014-05-13 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Optical character recognition system, aka OCR, allows identifying and recognizing printed characters from pictures. A wide range of devices already has such functionality, e.g, scanners and mobile devices. The current everyday tasks has an increasing demand for handwritten character recognition, for example, recognize specified amount on bank checks, identify postal address to automate some aspects of letter delivery. Handwritten digit recognition faces the difficulty of great intraclass variability, due to different writing stiles and different character slant degrees. This work presents three strategies to address handwritten digit recognition by means of three pattern recognition methods and two feature extraction algorithms. The first strategy makes use of Fourier Descriptor and Boundary Transition Technique to extract representative values from digits contours in order to recognize digits is used a neural network Multilayer Perceptron and a set of Support Vector Machines classifiers to validate neural network output. The second strategy represents this work's baseline using the classic convolutional neural networks algorithm from literature, LeNet5. Such algorithm received as input the raw digit images without preprocessing. The third strategy used a unbalanced decision tree in which support vector machines actuated as decision points and as representative feature received the raw digit images. Late experiments showed that first strategy was not effective enough to recognize digits; only about 80% of characters were successfully recognized. By means of Convolutional Neural Network was possible to achieve 0.9% of error rate, not so impressive if compared to literature best results. The third strategy was capable to recognize 100% of test samples from handwritten digits dataset of MNist. Each support vector machine classifier achieved 0% of error rate, due to an enormous amount of support vectors. / Sistemas de reconhecimento ótico de caracteres, também conhecidos como OCR, permitem identificar e reconhecer caracteres impressos por meio de imagens, uma funcionalidade já bem difundida em scanners, dispositivos móveis, entre outros. Existe uma crescente necessidade de reconhecimento de caracteres manuscritos para uso em várias situações, tais como reconhecimento de valores nominais em cheques de bancos, reconhecimento dos dígitos manuscritos de endereço postal para redirecionamento automatizado de cartas nos correios. Reconhecimento de dígitos manuscritos esbarra na dificuldade de lidar com uma grande variação intraclasse, devido a diferentes estilos de escrita, diferentes graus de inclinação dos caracteres. Este trabalho apresenta três estratégias utilizando três diferentes métodos de reconhecimento de padrões e dois métodos de extração de características. A primeira estratégia utilizou Descritores de Fourier e a técnica de transição de borda para extrair valores representativos do contorno dos caracteres e como camada de classificação utilizou uma rede neural MLP em associação com um conjunto de classificadores SVM para validar e corrigir eventuais erros da rede MLP. A segunda estratégia figurou como base comparativa para as demais estratégias por utilizar um algoritmo clássico de redes neurais convolutivas, LeNet5, e como características utilizou as próprias imagens dos dígitos. A terceira estratégia fez uso de um conjunto de classificadores SVM em uma árvore de decisão desbalanceada para a classificação dos dígitos a partir unicamente de suas imagens. Como resultados dos experimentos, a primeira estratégia provou não ser totalmente efetiva por obter resultados em torno de 80% de taxa de acerto. A segunda estratégia obteve 0,9% de taxa de erro que apesar de ter sido alta, ainda é muito menor se comparada com os melhores resultados obtidos na literatura. A terceira estratégia por sua vez logrou sucesso em reconhecer 100% das amostras de teste da base MNist de dígitos manuscritos, devido ao sucesso do treinamento de cada um dos classificadores SVM, que apesar de utilizarem uma enorme quantidade de vetores de suporte, atingiram individualmente 0% de taxa de erro.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-05-30T20:07:56Z No. of bitstreams: 1 robertocarlossoaresnalonpereirasouza.pdf: 1346845 bytes, checksum: e248f967f42f4ef763b613dc39ed0649 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-06-01T11:51:04Z (GMT) No. of bitstreams: 1 robertocarlossoaresnalonpereirasouza.pdf: 1346845 bytes, checksum: e248f967f42f4ef763b613dc39ed0649 (MD5) / Made available in DSpace on 2017-06-01T11:51:04Z (GMT). No. of bitstreams: 1 robertocarlossoaresnalonpereirasouza.pdf: 1346845 bytes, checksum: e248f967f42f4ef763b613dc39ed0649 (MD5) Previous issue date: 2013-02-21 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Neste trabalho apresenta-se uma nova formulação para regressão ortogonal. O problema é definido como a minimização do risco empírico em relação a uma função de perda com tubo desenvolvida para regressão ortogonal, chamada ρ-insensível. Um algoritmo para resolver esse problema é proposto, baseado na abordagem da descida do gradiente estocástica. Quando formulado em variáveis duais o método permite a introdução de funções kernel e flexibilidade do tubo. Até onde se sabe, este é o primeiro método que permite a introdução de kernels, através do chamado “kernel-trick”, para regressão ortogonal. Apresenta-se ainda um algoritmo para regressão clássica que usa a função de perda ε-insensível e segue também a abordagem da descida do gradiente. Para esse algo ritmo apresenta-se uma prova de convergência que garante um número finito de correções. Finalmente, introduz-se uma estratégia incremental que pode ser usada acoplada com ambos os algoritmos para obter soluções esparsas e também uma aproximação para o “tubo mínimo”que contém os dados. Experimentos numéricos são apresentados e os resultados comparados a outros métodos da literatura. / In this work, we introduce a new formulation for orthogonal regression. The problem is defined as minimization of the empirical risk with respect to a tube loss function de veloped for orthogonal regression, named ρ-insensitive. The method is constructed via an stochastic gradient descent approach. The algorithm can be used in primal or in dual variables. The latter formulation allows the introduction of kernels and soft margins. To the best of our knowledge, this is the first method that allows the introduction of kernels via the so-called “kernel-trick” for orthogonal regression. Also, we present an algorithm to solve the classical regression problem using the ε-insensitive loss function. A conver gence proof that guarantees a finite number of updates is presented for this algorithm. In addition, an incremental strategy algorithm is introduced, which can be used to find sparse solutions and also an approximation to the “minimal tube” containing the data. Numerical experiments are shown and the results compared with other methods.

Orientador: Robson Pederiva / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-19T20:21:22Z (GMT). No. of bitstreams: 1 Silva_ViniciusAugustoDiniz_M.pdf: 16029999 bytes, checksum: a3585cbd021c6f84637d409a34a51962 (MD5) Previous issue date: 2012 / Resumo: Motores elétricos são componentes essenciais na grande maioria dos processos industriais. As diversas falhas nas máquinas de indução podem gerar consequências drásticas para um processo industrial. Os principais problemas estão relacionados ao aumento dos custos, piora nas condições do processo e de segurança e qualidade do produto final. Muitas destas falhas mostram-se progressivas. Neste trabalho, apresenta-se uma contribuição ao estudo de métodos de detecção de falhas em motores elétricos usando Máquinas de Vetores de Suporte (SVM), treinadas a partir de sinais de vibração obtidos experimentalmente. A metodologia desenvolvida é usada para classificar a excitação devido a falhas mecânicas e elétricas, além da condição normal de funcionamento, utilizando apenas um sensor de vibração. Através da seleção de parâmetros é possível reduzir o número de entradas capazes de representar os sinais utilizados para o treinamento das SVMs. A normalização proposta permitiu melhorar as taxas de acerto, quando se quer classificar falhas em diferentes níveis de severidade das que foram utilizadas para o treinamento. Os resultados mostraram que a metodologia apresentada pode ser adaptada para ser utilizada em aplicações práticas industriais e poderá ser no futuro uma saída viável para uma manutenção industrial eficiente e eficaz / Abstract: Electric motors are essential components in most industrial processes. The several faults in induction machines can produce drastic consequences for an industrial process. The main problems are related to rising costs, decrease conditions in the process and safety and quality of the final product. Many of these failures are progressive. In this paper, we present a contribution to the study of methods for detecting faults in induction motors using Support Vector Machines (SVM) trained from vibration signals obtained experimentally. The developed methodology is used to classify the excitation due to mechanical and electrical failures, in addition to normal operating condition, using only a vibration sensor. Through the feature selection is possible to reduce the number of inputs that represent the signals used for training the SVMs. The proposed standardization has improved the accuracy rates when we want to classify failures at different levels of severity that were used for training. The results showed that this methodology can be adapted for use in industrial and practical applications and in the future may be a viable approach to an efficient and effective industrial maintenance / Mestrado / Mecanica dos Sólidos e Projeto Mecanico / Mestre em Engenharia Mecânica

L’estimation de la pose du corps humain est un problème difficile en vision par ordinateur et les actions de toutes les difficultés de détection d’objet. Cette thèse se concentre sur les problèmes de l’estimation de la pose du corps humain dans les images ou vidéo, y compris la diversité des apparences, les changements de scène et l’éclairage de fond de confusion encombrement. Pour résoudre ces problèmes, nous construisons un modèle robuste comprenant les éléments suivants. Tout d’abord, les méthodes top-down et bottom-up sont combinés à l’estimation pose humaine. Nous étendons le modèle structure picturale (PS) de coopérer avec filtre à particules recuit (APF) pour robuste multi-vues estimation de la pose. Deuxièmement, nous proposons plusieurs parties de mélange à base (MMP) modèle d’une partie supérieure du corps pour l’estimation de la pose qui contient deux étapes. Dans la phase de pré-estimation, il y a trois étapes: la détection du haut du corps, catégorie estimation du modèle pour le haut du corps, et la sélection de modèle complet pour pose estimation. Dans l’étape de l’estimation, nous abordons le problème d’une variété de poses et les activités humaines. Enfin, le réseau de neurones à convolution (CNN) est introduit pour l’estimation de la pose. Un Local Multi-résolution réseau de neurones à convolution (LMR-CNN) est proposé pour apprendre la représentation pour chaque partie du corps. En outre, un modèle hiérarchique sur la base LMR-CNN est défini pour faire face à la complexité structurelle des parties de branche. Les résultats expérimentaux démontrent l’efficacité du modèle proposé / Human pose estimation is a challenging problem in computer vision and shares all the difficulties of object detection. This thesis focuses on the problems of human pose estimation in still images or video, including the diversity of appearances, changes in scene illumination and confounding background clutter. To tackle these problems, we build a robust model consisting of the following components. First, the top-down and bottom-up methods are combined to estimation human pose. We extend the Pictorial Structure (PS) model to cooperate with annealed particle filter (APF) for robust multi-view pose estimation. Second, we propose an upper body based multiple mixture parts (MMP) model for human pose estimation that contains two stages. In the pre-estimation stage, there are three steps: upper body detection, model category estimation for upper body, and full model selection for pose estimation. In the estimation stage, we address the problem of a variety of human poses and activities. Finally, a Deep Convolutional Neural Network (DCNN) is introduced for human pose estimation. A Local Multi-Resolution Convolutional Neural Network (LMR-CNN) is proposed to learn the representation for each body part. Moreover, a LMR-CNN based hierarchical model is defined to meet the structural complexity of limb parts. The experimental results demonstrate the effectiveness of the proposed model

The paper describes predictive modelling of biomarker data stemming from patients suffering from multiple sclerosis. Improvements of multivariate analyses of the data are investigated with the goal of increasing the capability to assign samples to correct subgroups from the data alone. The effects of different preceding scalings of the data are investigated and combinations of multivariate modelling methods and variable selection methods are evaluated. Attempts at merging the predictive capabilities of the method combinations through voting-procedures are made. A technique for improving the result of PLS-modelling, called bagging, is evaluated. The best methods of multivariate analysis of the ones tried are found to be Partial least squares (PLS) and Support vector machines (SVM). It is concluded that the scaling have little effect on the prediction performance for most methods. The method combinations have interesting properties – the default variable selections of the multivariate methods are not always the best. Bagging improves performance, but at a high cost. No reasons for drastically changing the work flows of the biomarker data analysis are found, but slight improvements are possible. Further research is needed.

In this work we compare the prediction performance of three optimized technical indicators with a Support Vector Machine Neural Network. For the indicator part we picked the common used indicators: Relative Strength Index, Moving Average Convergence Divergence and Stochastic Oscillator. For the Support Vector Machine we used a radial-basis kernel function and regression mode. The techniques were applied on financial time series brought from the Swedish stock market. The comparison and the promising results should be of interest for both finance people using the techniques in practice, as well as software companies and similar considering to implement the techniques in their products.

Current Ambient Intelligence and Intelligent Environment research focuses on the interpretation of a subject’s behaviour at the activity level by logging the Activity of Daily Living (ADL) such as eating, cooking, etc. In general, the sensors employed (e.g. PIR sensors, contact sensors) provide low resolution information. Meanwhile, the expansion of ubiquitous computing allows researchers to gather additional information from different types of sensor which is possible to improve activity analysis. Based on the previous research about sitting posture detection, this research attempts to further analyses human sitting activity. The aim of this research is to use non-intrusive low cost pressure sensor embedded chair system to recognize a subject’s activity by using their detected postures. There are three steps for this research, the first step is to find a hardware solution for low cost sitting posture detection, second step is to find a suitable strategy of sitting posture detection and the last step is to correlate the time-ordered sitting posture sequences with sitting activity. The author initiated a prototype type of sensing system called IntelliChair for sitting posture detection. Two experiments are proceeded in order to determine the hardware architecture of IntelliChair system. The prototype looks at the sensor selection and integration of various sensor and indicates the best for a low cost, non-intrusive system. Subsequently, this research implements signal process theory to explore the frequency feature of sitting posture, for the purpose of determining a suitable sampling rate for IntelliChair system. For second and third step, ten subjects are recruited for the sitting posture data and sitting activity data collection. The former dataset is collected byasking subjects to perform certain pre-defined sitting postures on IntelliChair and it is used for posture recognition experiment. The latter dataset is collected by asking the subjects to perform their normal sitting activity routine on IntelliChair for four hours, and the dataset is used for activity modelling and recognition experiment. For the posture recognition experiment, two Support Vector Machine (SVM) based classifiers are trained (one for spine postures and the other one for leg postures), and their performance evaluated. Hidden Markov Model is utilized for sitting activity modelling and recognition in order to establish the selected sitting activities from sitting posture sequences.2. After experimenting with possible sensors, Force Sensing Resistor (FSR) is selected as the pressure sensing unit for IntelliChair. Eight FSRs are mounted on the seat and back of a chair to gather haptic (i.e., touch-based) posture information. Furthermore, the research explores the possibility of using alternative non-intrusive sensing technology (i.e. vision based Kinect Sensor from Microsoft) and find out the Kinect sensor is not reliable for sitting posture detection due to the joint drifting problem. A suitable sampling rate for IntelliChair is determined according to the experiment result which is 6 Hz. The posture classification performance shows that the SVM based classifier is robust to “familiar” subject data (accuracy is 99.8% with spine postures and 99.9% with leg postures). When dealing with “unfamiliar” subject data, the accuracy is 80.7% for spine posture classification and 42.3% for leg posture classification. The result of activity recognition achieves 41.27% accuracy among four selected activities (i.e. relax, play game, working with PC and watching video). The result of this thesis shows that different individual body characteristics and sitting habits influence both sitting posture and sitting activity recognition. In this case, it suggests that IntelliChair is suitable for individual usage but a training stage is required.

Machine learning (ML) methods have been applied to the analysis of a range of biological systems. This thesis evaluates the application of these methods to the problem domain of skin permeability. ML methods offer great potential in both predictive ability and their ability to provide mechanistic insight to, in this case, the phenomena of skin permeation. Historically, refining mathematical models used to predict percutaneous drug absorption has been thought of as a key factor in this field. Quantitative Structure-Activity Relationships (QSARs) models are used extensively for this purpose. However, advanced ML methods successfully outperform the traditional linear QSAR models. In this thesis, the application of ML methods to percutaneous absorption are investigated and evaluated. The major approach used in this thesis is Gaussian process (GP) regression method. This research seeks to enhance the prediction performance by using local non-linear models obtained from applying clustering algorithms. In addition, to increase the model's quality, a kernel is generated based on both numerical chemical variables and categorical experimental descriptors. Monte Carlo algorithm is also employed to generate reliable models from variable data which is inevitable in biological experiments. The datasets used for this study are small and it may raise the over-fitting/under-fitting problem. In this research I attempt to find optimal values of skin permeability using GP optimisation algorithms within small datasets. Although these methods are applied here to the field of percutaneous absorption, it may be applied more broadly to any biological system.

Magister Scientiae - MSc / We investigate the use of noise degraded facial images in the application of facial expression recognition. In particular, we trained Gabor+SVMclassifiers to recognize facial expressions images with various types of noise. We applied Gaussian noise, Poisson noise, varying levels of salt and pepper noise, and speckle noise to noiseless facial images. Classifiers were trained with images without noise and then tested on the images with noise. Next, the classifiers were trained using images with noise, and then on tested both images that had noise, and images that were noiseless. Finally, classifiers were tested on images while increasing the levels of salt and pepper in the test set. Our results reflected distinct degradation of recognition accuracy. We also discovered that certain types of noise, particularly Gaussian and Poisson noise, boost recognition rates to levels greater than would be achieved by normal, noiseless images. We attribute this effect to the Gaussian envelope component of Gabor filters being sympathetic to Gaussian-like noise, which is similar in variance to that of the Gabor filters. Finally, using linear regression, we mapped a mathematical model to this degradation and used it to suggest how recognition rates would degrade further should more noise be added to the images. / South Africa