Thesis (PhD (Process Engineering))--University of Stellenbosch, 2010. / Dissertation presented for the Degree of DOCTOR OF PHILOSOPHY (Extractive Metallurgical Engineering) in the Department of Process Engineering at the University of Stellenbosch / ENGLISH ABSTRACT: Fault diagnosis is an important component of process monitoring, relevant in the greater context of developing safer, cleaner and more cost efficient processes. Data-driven unsupervised (or feature extractive) approaches to fault diagnosis exploit the many measurements available on modern plants. Certain current unsupervised approaches are hampered by their linearity assumptions, motivating the investigation of nonlinear methods. The diversity of data structures also motivates the investigation of novel feature extraction methodologies in process monitoring. Random forests are recently proposed statistical inference tools, deriving their predictive accuracy from the nonlinear nature of their constituent decision tree members and the power of ensembles. Random forest committees provide more than just predictions; model information on data proximities can be exploited to provide random forest features. Variable importance measures show which variables are closely associated with a chosen response variable, while partial dependencies indicate the relation of important variables to said response variable. The purpose of this study was therefore to investigate the feasibility of a new unsupervised method based on random forests as a potentially viable contender in the process monitoring statistical tool family. The hypothesis investigated was that unsupervised process monitoring and fault diagnosis can be improved by using features extracted from data with random forests, with further interpretation of fault conditions aided by random forest tools. The experimental results presented in this work support this hypothesis. An initial study was performed to assess the quality of random forest features. Random forest features were shown to be generally difficult to interpret in terms of geometry present in the original variable space. Random forest mapping and demapping models were shown to be very accurate on training data, and to extrapolate weakly to unseen data that do not fall within regions populated by training data. Random forest feature extraction was applied to unsupervised fault diagnosis for process data, and compared to linear and nonlinear methods. Random forest results were comparable to existing techniques, with the majority of random forest detections due to variable reconstruction errors. Further investigation revealed that the residual detection success of random forests originates from the constrained responses and poor generalization artifacts of decision trees. Random forest variable importance measures and partial dependencies were incorporated in a visualization tool to allow for the interpretation of fault conditions. A dynamic change point detection application with random forests proved more successful than an existing principal component analysis-based approach, with the success of the random forest method again residing in reconstruction errors. The addition of random forest fault diagnosis and change point detection algorithms to a suite of abnormal event detection techniques is recommended. The distance-to-model diagnostic based on random forest mapping and demapping proved successful in this work, and the theoretical understanding gained supports the application of this method to further data sets. / AFRIKAANSE OPSOMMING: Foutdiagnose is ’n belangrike komponent van prosesmonitering, en is relevant binne die groter konteks van die ontwikkeling van veiliger, skoner en meer koste-effektiewe prosesse. Data-gedrewe toesigvrye of kenmerkekstraksie-benaderings tot foutdiagnose benut die vele metings wat op moderne prosesaanlegte beskikbaar is. Party van die huidige toesigvrye benaderings word deur aannames rakende liniariteit belemmer, wat as motivering dien om nie-liniêre metodes te ondersoek. Die diversiteit van datastrukture is ook verdere motivering vir ondersoek na nuwe kenmerkekstraksiemetodes in prosesmonitering. Lukrake-woude is ’n nuwe statistiese inferensie-tegniek, waarvan die akkuraatheid toegeskryf kan word aan die nie-liniêre aard van besluitnemingsboomlede en die bekwaamheid van ensembles. Lukrake-woudkomitees verskaf meer as net voorspellings; modelinligting oor datapuntnabyheid kan benut word om lukrakewoudkenmerke te verskaf. Metingbelangrikheidsaanduiers wys watter metings in ’n noue verhouding met ’n gekose uitsetveranderlike verkeer, terwyl parsiële afhanklikhede aandui wat die verhouding van ’n belangrike meting tot die gekose uitsetveranderlike is. Die doel van hierdie studie was dus om die uitvoerbaarheid van ’n nuwe toesigvrye metode vir prosesmonitering gebaseer op lukrake-woude te ondersoek. Die ondersoekte hipotese lui: toesigvrye prosesmonitering en foutdiagnose kan verbeter word deur kenmerke te gebruik wat met lukrake-woude geëkstraheer is, waar die verdere interpretasie van foutkondisies deur addisionele lukrake-woude-tegnieke bygestaan word. Eksperimentele resultate wat in hierdie werkstuk voorgelê is, ondersteun hierdie hipotese. ’n Intreestudie is gedoen om die gehalte van lukrake-woudkenmerke te assesseer. Daar is bevind dat dit moeilik is om lukrake-woudkenmerke in terme van die geometrie van die oorspronklike metingspasie te interpreteer. Verder is daar bevind dat lukrake-woudkartering en -dekartering baie akkuraat is vir opleidingsdata, maar dat dit swak ekstrapolasie-eienskappe toon vir ongesiene data wat in gebiede buite dié van die opleidingsdata val. Lukrake-woudkenmerkekstraksie is in toesigvrye-foutdiagnose vir gestadigde-toestandprosesse toegepas, en is met liniêre en nie-liniêre metodes vergelyk. Resultate met lukrake-woude is vergelykbaar met dié van bestaande metodes, en die meerderheid lukrake-woudopsporings is aan metingrekonstruksiefoute toe te skryf. Verdere ondersoek het getoon dat die sukses van res-opsporing op die beperkte uitsetwaardes en swak veralgemenende eienskappe van besluitnemingsbome berus. Lukrake-woude-metingbelangrikheidsaanduiers en parsiële afhanklikhede is ingelyf in ’n visualiseringstegniek wat vir die interpretasie van foutkondisies voorsiening maak. ’n Dinamiese aanwending van veranderingspuntopsporing met lukrake-woude is as meer suksesvol bewys as ’n bestaande metode gebaseer op hoofkomponentanalise. Die sukses van die lukrake-woudmetode is weereens aan rekonstruksie-reswaardes toe te skryf. ’n Voorstel wat na aanleiding van hierde studie gemaak is, is dat die lukrake-woudveranderingspunt- en foutopsporingsmetodes by ’n soortgelyke stel metodes gevoeg kan word. Daar is in hierdie werk bevind dat die afstand-vanaf-modeldiagnostiek gebaseer op lukrake-woudkartering en -dekartering suksesvol is vir foutopsporing. Die teoretiese begrippe wat ontsluier is, ondersteun die toepassing van hierdie metodes op verdere datastelle.

The proposed study develops a framework that can accurately capture and model input and output variables for multidisciplinary systems to mitigate the computational cost when uncertainties are involved. The dimension of the random input variables is reduced depending on the degree of correlation calculated by relative entropy. Feature extraction methods; namely Principal Component Analysis (PCA), the Auto-Encoder (AE) algorithm are developed when the input variables are highly correlated. The Independent Features Test (IndFeaT) is implemented as the feature selection method if the correlation is low to select a critical subset of model features. Moreover, Artificial Neural Network (ANN) including Probabilistic Neural Network (PNN) is integrated into the framework to correctly capture the complex response behavior of the multidisciplinary system with low computational cost. The efficacy of the proposed method is demonstrated with electro-mechanical engineering examples including a solder joint and stretchable patch antenna examples.

A new method of registering multiple range datasets collected in a GPS-denied, tunnel-like environment is presented. The method is designed to function with minimal user inputs and be effective over a wide range of changes in observation angle. The method is initially developed to operate on data in a general 2.5D coordinate system. Then, the general registration method is specifically tailored to a 2.5D spherical coordinate system. To apply the method, the range data is first filtered with a series of discrete Gaussian-based filters to construct a second-order Taylor series approximation to the surface about each sampled point. Finally, principal curvatures are calculated and compared across neighboring datasets to determine homologies and the best fit transfer matrix. The new method relaxes the minimum change in perspective requirement between neighboring datasets typical of other algorithms. Results from the application of the method on both synthetic and real-world data are shown. The real-world data comes from a series of high explosive tests performed in a tunnel environment. The tunnels were oriented horizontally in rock and constructed with boring equipment. The tunnel surfaces were surveyed with a Faro Focus3D terrestrial panorama scanning light detection and ranging (lidar) system both before and after a high explosive device was detonated inside the tunnel with the intent of documenting damage to the tunnel surface.

The design of a machine which reads unconstrained words still remains an unsolved problem. For example, automatic interpretation of handwritten documents by a computer is still under research. Most systems attempt to segment words into letters and read words one character at a time. However, segmenting handwritten words is very difficult. So to avoid this words are treated as a whole. This research investigates a number of features computed from whole words for the recognition of handwritten words in particular. Arabic text classification and recognition is a complicated process compared to Latin and Chinese text recognition systems. This is due to the nature cursiveness of Arabic text. The work presented in this thesis is proposed for word based recognition of handwritten Arabic scripts. This work is divided into three main stages to provide a recognition system. The first stage is the pre-processing, which applies efficient pre-processing methods which are essential for automatic recognition of handwritten documents. In this stage, techniques for detecting baseline and segmenting words in handwritten Arabic text are presented. Then connected components are extracted, and distances between different components are analyzed. The statistical distribution of these distances is then obtained to determine an optimal threshold for word segmentation. The second stage is feature extraction. This stage makes use of the normalized images to extract features that are essential in recognizing the images. Various method of feature extraction are implemented and examined. The third and final stage is the classification. Various classifiers are used for classification such as K nearest neighbour classifier (k-NN), neural network classifier (NN), Hidden Markov models (HMMs), and the Dynamic Bayesian Network (DBN). To test this concept, the particular pattern recognition problem studied is the classification of 32492 words using ii the IFN/ENIT database. The results were promising and very encouraging in terms of improved baseline detection and word segmentation for further recognition. Moreover, several feature subsets were examined and a best recognition performance of 81.5% is achieved.

Mammography is used to aid early detection and diagnosis systems. It takes an x-ray image of the breast and can provide a second opinion for radiologists. The earlier detection is made, the better treatment works. Digital mammograms are dealt with by Computer Aided Diagnosis (CAD) systems that can detect and analyze abnormalities in a mammogram. The purpose of this study is to investigate how to categories cropped regions of interest (ROI) from digital mammogram images into two classes; normal and abnormal regions (which contain microcalcifications). The work proposed in this thesis is divided into three stages to provide a concept system for classification between normal and abnormal cases. The first stage is the Segmentation Process, which applies thresholding filters to separate the abnormal objects (foreground) from the breast tissue (background). Moreover, this study has been carried out on mammogram images and mainly on cropped ROI images from different sizes that represent individual microcalcification and ROI that represent a cluster of microcalcifications. The second stage in this thesis is feature extraction. This stage makes use of the segmented ROI images to extract characteristic features that would help in identifying regions of interest. The wavelet transform has been utilized for this process as it provides a variety of features that could be examined in future studies. The third and final stage is classification, where machine learning is applied to be able to distinguish between normal ROI images and ROI images that may contain microcalcifications. The result indicated was that by combining wavelet transform and SVM we can distinguish between regions with normal breast tissue and regions that include microcalcifications.

Time series data mining is one branch of data mining. Time series analysis and prediction have always played an important role in human activities and natural sciences. A Pseudo-Periodical time series has a complex structure, with fluctuations and frequencies of the times series changing over time. Currently, Pseudo-Periodicity of time series brings new properties and challenges to time series analysis and prediction. This thesis proposes two original computational approaches for time series analysis and prediction: Moving Average of nth-order Difference (MANoD) and Series Features Extraction (SFE). Based on data-driven methods, the two original approaches open new insights in time series analysis and prediction contributing with new feature detection techniques. The proposed algorithms can reveal hidden patterns based on the characteristics of time series, and they can be applied for predicting forthcoming events. This thesis also presents the evaluation results of proposed algorithms on various pseudo-periodical time series, and compares the predicting results with classical time series prediction methods. The results of the original approaches applied to real world and synthetic time series are very good and show that the contributions open promising research directions.

At its core, a nanopore detector has a nanometer-scale biological membrane across which a voltage is applied. The voltage draws a DNA molecule into an á-hemolysin channel in the membrane. Consequently, a distinctive channel current blockade signal is created as the molecule flexes and interacts with the channel. This flexing of the molecule is characterized by different blockade levels in the channel current signal. Previous experiments have shown that a nanopore detector is sufficiently sensitive such that nearly identical DNA molecules were classified successfully using machine learning techniques such as Hidden Markov Models and Support Vector Machines in a channel current based signal analysis platform [4-9]. In this paper, methods for improving feature extraction are presented to improve both classification and to provide biologists and chemists with a better understanding of the physical properties of a given molecule.

Behavioural biometrics are unique non-physical human characteristics that can be used to distinguish one person from another. One such characteristic, which belongs to the Gait Biometric, is the footstep Ground Reaction Force (GRF), the temporal signature of the force exerted by the ground back on the foot through the course of a footstep. This is a biometric for which the computational power required for practical applications in a security setting has only recently become available. In spite of this, there are still barriers to deployment in a practical setting, including large research gaps concerning the effect of footwear and stepping speed on footstep GRF-based person recognition. In this thesis we devised an experiment to address these research gaps, while also expanding upon the biometric system research presented in previous GRF recognition studies. To assess the effect of footwear on recognition performance we proposed the analysis of a dataset containing samples for two different types of running shoes. While, with regards to stepping speed, we set out to demonstrate that normalizing for step duration will mitigate speed variation biases and improve GRF recognition performance; this included the development of two novel machine learning-based temporal normalization techniques: Localized Least Squares Regression (LLSR) and Localized Least Squares Regression with Dynamic Time Warping (LLSRDTW). Moreover, building upon previous research, biometric system analysis was done over four feature extractors, seven normalizers, and five different classifiers, allowing us to indirectly compare the GRF recognition results for biometric system configurations that had never before been directly compared. The results achieved for the aforementioned experiment were generally in line with our initial assumptions. Comparing biometrics systems trained and tested with the same footwear against those trained and tested with different footwear, we found an average decrease in recognition performance of about 50%. While, performing LLSRDTW step duration normalization on the data led to a 14-15% improvement in recognition performance over its non-normalized equivalent in our two most stable feature spaces. Examining our biometric system configurations we found that a Wavelet Packet Decomposition-based feature extractor produced our best feature space results with an EER average of about 2.6%, while the Linear Discriminant Analysis (LDA) classifier performed best of the classifiers, about 19% better than any of the others. Finally, while not the intended purpose of our research, the work in this thesis was presented such that it may form a foundation upon which future classification problems could be approached in a wide range of alternative domains. / Graduate / 0800 / 0544 / jericmason@gmail.com

Imagens multiespectrais estão presentes na grande maioria de dispositivos de imageamento atuais, desde câmeras pessoais até microscópios, telescópios e satélites. No entanto, grande parte dos trabalhos em análise de texturas e afins propõem abordagens monocromáticas, que muitas vezes consideram apenas níveis de cinza. Nesse contexto e considerando o aumento da capacidade dos computadores atuais, o uso da informação espectral deve ser considerada na construção de modelos melhores. Ultimamente redes neurais convolucionais profundas pré-treinadas tem sido usadas em imagens coloridas de 3 canais, porém são limitadas a apenas esse formato e computam muitas convoluções, o que demanda por hardware específico (GPU). Esses fatos motivaram esse trabalho, que propõem técnicas para a modelagem e caracterização de imagens multiespectrais baseadas em redes complexas, que tem se mostrado uma ferramenta eficiente em trabalhos anteriores e possui complexidade computacional similar à métodos tradicionais. São introduzidas duas abordagens para aplicação em imagens coloridas de três canais, denominadas Rede Multicamada (RM) e Rede Multicamada Direcionada (RMD). Esses métodos modelam todos os canais da imagem de forma conjunta, onde as redes possuem conexões intra e entre canais, de forma parecida ao processamento oponente de cor do sistema visual humano. Experimentos em cinco bases de textura colorida mostram a proposta RMD supera vários métodos da literatura no geral, incluindo redes convolucionais e métodos tradicionais integrativos. Além disso, as propostas demonstraram alta robustez a diferentes espaços de cor (RGB, LAB, HSV e I1I2I3), enquanto que outros métodos oscilam de base para base. Também é proposto um método para caracterizar imagens multiespectrais de muitos canais, denominado Rede Direcionada de Similaridade Angular (RDSA). Nessa proposta, cada pixel multiespectral é considerado como um vetor de dimensão equivalente à quantidade de canais da imagem e o peso das arestas representa sua similaridade do cosseno, apontando para o pixel de maior valor absoluto. Esse método é aplicado em um conjunto de imagens de microscopia por fluorescência de 32 canais, em um experimento para identificar variações na estrutura foliar do espécime Jacaranda Caroba submetidos à diferentes condições. O método RDSA obtém as maiores taxas de acerto de classificação nesse conjunto de dados, com 91, 9% usando o esquema de validação cruzada Leave-one-out e 90, 5(±1, 1)% com 10-pastas, contra 81, 8% e 84, 7(±2, 2) da rede convolucional VGG16. / Multispectral images are present in the vast majority of current imaging devices, from personal cameras to microscopes, telescopes and satellites. However, much of the work in texture analysis and the like proposes monochromatic approaches, which often consider only gray levels. In this context and considering the performance increase of current computers, the use of the spectral information must be considered in the construction of better models. Lately, pre-trained deep convolutional neural networks have been used in 3-channel color images, however they are limited to just this format and compute many convolutions, which demands specific hardware (GPU). These facts motivated this work, which propose techniques for the modeling and characterization of multispectral images based on complex networks, which has proved to be an efficient tool in previous works and has computational complexity similar to traditional methods. Two approaches are introduced for application in 3-channel color images, called Multilayer Network (RM) and Directed Multilayer Network (RMD). These methods model all channels of the image together, where the networks have intra- and inter-channel connections, similar to the opponent color processing of the human visual system. Experiments in five color texture datasets shows that the RMD proposal overcomes several methods of the literature in general, including convolutional networks and traditional integrative methods. In addition, the proposals have demonstrated high robustness to different color spaces (RGB, LAB, HSV and I1I2I3), while other methods oscillate from dataset to dataset. Moreover it is proposed a new method to characterize multispectral images of many channels, called Directed Network of Angular Similarity (RDSA). In this proposal, each multispectral pixel is considered as a vector of dimensions equivalent to the number of channels of the image and the weight of the edges represents its cosine similarity, pointing to the pixel of greatest absolute value. This method is applied to a set of fluorescence microscopy images of 32 channels in an experiment to identify variations in the leaf structure of the Jacaranda Caroba specimen under different conditions. The RDSA method obtains the highest classification rates in this dataset, with 91.9% with the Leave-one-out cross-validation scheme and 90.5(±1.1)% with 10-folds, against 81.8% and 84.7(±2.2) of the convolutional network VGG16.

Obter uma representação sucinta e representativa de imagens médicas é um desafio que tem sido perseguido por pesquisadores da área de processamento de imagens médicas com o propósito de apoiar o diagnóstico auxiliado por computador (Computer Aided Diagnosis - CAD). Os sistemas CAD utilizam algoritmos de extração de características para representar imagens, assim, diferentes extratores podem ser avaliados. No entanto, as imagens médicas contêm estruturas internas que são importantes para a identificação de tecidos, órgãos, malformações ou doenças. É usual que um grande número de características sejam extraídas das imagens, porém esse fato que poderia ser benéfico, pode na realidade prejudicar o processo de indexação e recuperação das imagens com problemas como a maldição da dimensionalidade. Assim, precisa-se selecionar as características mais relevantes para tornar o processo mais eficiente e eficaz. Esse trabalho desenvolveu o método de seleção supervisionada de características FSCoMS (Feature Selection based on Compactness Measure from Scatterplots) para obter o ranking das características, contemplando assim, o que é necessário para o tipo de imagens médicas sob análise. Dessa forma, produziu-se vetores de características mais enxutos e eficientes para responder consultas por similaridade. Adicionalmente, foi desenvolvido o extrator de características k-Gabor que extrai características por níveis de cinza, ressaltando estruturas internas das imagens médicas. Os experimentos realizados foram feitos com quatro bases de imagens médicas do mundo real, onde o k-Gabor sobressai pelo desempenho na recuperação por similaridade de imagens médicas, enquanto o FSCoMS reduz a redundância das características para obter um vetor de características menor do que os métodos de seleção de características convencionais e ainda com um maior desempenho em recuperação de imagens / Obtaining a representative and succinct description of medical images is a challenge that has been pursued by researchers in the area of medical image processing to support Computer-Aided Diagnosis (CAD). CAD systems use feature extraction algorithms to represent images. Thus, different extractors can be evaluated. However, medical images contain important internal structures that allow identifying tissues, organs, deformations and diseases. It is usual that a large number of features are extracted the images. Nevertheless, what appears to be beneficial actually impairs the process of indexing and retrieval of images, revealing problems such as the curse of dimensionality. Thus, it is necessary to select the most relevant features to make the process more efficient and effective. This dissertation developed a supervised feature selection method called FSCoMS (Feature Selection based on Compactness Measure from Scatterplots) in order to obtain a ranking of features, suitable for medical image analysis. Our method FSCoMS had generated shorter and efficient feature vectors to answer similarity queries. Additionally, the k-Gabor feature extractor was developed, which extracts features by gray levels, highlighting internal structures of medical images. The experiments performed were performed on four real world medical datasets. Results have shown that the k-Gabor boosts the retrieval performance, whereas the FSCoMS reduces the subsets redundancy to produce a more compact feature vector than the conventional feature selection methods and even with a higher performance in image retrieval