Nous présentons deux contributions majeures : 1) une combinaison de plusieurs descripteurs d’images pour la classification à grande échelle, 2) des algorithmes parallèles de SVM pour la classification d’images à grande échelle. Nous proposons aussi un algorithme incrémental et parallèle de classification lorsque les données ne peuvent plus tenir en mémoire vive. / We have proposed a novel method of combination multiple of different features for image classification. For large scale learning classifiers, we have developed the parallel versions of both state-of-the-art linear and nonlinear SVMs. We have also proposed a novel algorithm to extend stochastic gradient descent SVM for large scale learning. A class of large scale incremental SVM classifiers has been developed in order to perform classification tasks on large datasets with very large number of classes and training data can not fit into memory.

Made available in DSpace on 2017-07-10T16:41:32Z (GMT). No. of bitstreams: 1 Dissertacao Altamira Queiroz2.pdf: 2299194 bytes, checksum: 07ecb127ebc11ad21a0cb551e23c2f1b (MD5) Previous issue date: 2016-02-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / One of the main problems currently faced by electric utilities is the occurrence of energy losses in the distribution network caused by fraud and electricity theft. Because of the financial losses and risks to public safety, the development of solutions to detect and combat fraud in the distribution networks is of the utmost importance. This work presents an analysis of computational intelligence algorithms to extract knowledge in databases with information from monthly energy consumption to identify consumption patterns with anomalies which could represent fraud. The algorithms Artificial Neural Networks and Support Vector Machines were tested to see which one perform better on the identification consumption patterns with abnormalities. Tests have shown that the algorithms used are able to detect patterns in electricity consumption curves, including special situations of fraud that manual techniques did not detect. / Um dos principais problemas que enfrentam atualmente as empresas concessionárias de energia elétrica é a ocorrência de perdas de energia na rede de distribuição, causadas por fraudes e furtos de energia elétrica. Sendo que tais problemas provocam prejuízos financeiros e também colocam em risco a segurança pública, é de grande interesse das concessionárias encontrar soluções para detectar e combater fraudes nas redes de distribuição de energia elétrica. Neste conceito, o presente trabalho apresenta uma análise dos algoritmos de Inteligência Computacional para extrair conhecimento de bases de dados de informações de consumo mensal de energia elétricas de usuários de uma determinada concessionária, a fim de identificar padrões de consumo com anomalias que representem possíveis fraudes nas redes de distribuição de energia elétrica. Para detectar padrões nas curvas de consumo, foram utilizados algoritmos de Redes Neurais Artificiais e Máquinas de Vetores de Suporte. Após a criação dos modelos, estes foram testados para verificar qual seria o melhor algoritmo para a detecção de padrões de consumo com anomalias, e os resultados obtidos, foram então, comparados com uma base de dados fornecida pela concessionária com a verificação manual dos usuários. Os testes demonstraram que os algoritmos utilizados são capazes de detectar padrões nas curvas de consumo de energia elétrica, inclusive detectando situações especiais de fraudes que técnicas manuais não detectaram.

Cette thèse s’inscrit dans le contexte de la reconnaissance des activités à partir de séquences vidéo qui est une des préoccupations majeures dans le domaine de la vision par ordinateur. Les domaines d'application pour ces systèmes de vision sont nombreux notamment la vidéo surveillance, la recherche et l'indexation automatique de vidéos ou encore l'assistance aux personnes âgées. Cette tâche reste problématique étant donnée les grandes variations dans la manière de réaliser les activités, l'apparence de la personne et les variations des conditions d'acquisition des activités. L'objectif principal de ce travail de thèse est de proposer une méthode de reconnaissance efficace par rapport aux différents facteurs de variabilité. Les représentations basées sur les points d'intérêt ont montré leur efficacité dans les travaux d'art; elles ont été généralement couplées avec des méthodes de classification globales vue que ses primitives sont temporellement et spatialement désordonnées. Les travaux les plus récents atteignent des performances élevées en modélisant le contexte spatio-temporel des points d'intérêts par exemple certains travaux encodent le voisinage des points d'intérêt à plusieurs échelles. Nous proposons une méthode de reconnaissance des activités qui modélise explicitement l'aspect séquentiel des activités tout en exploitant la robustesse des points d'intérêts dans les conditions réelles. Nous commençons par l'extractivité des points d'intérêt dont a montré leur robustesse par rapport à l'identité de la personne par une étude tensorielle. Ces primitives sont ensuite représentées en tant qu'une séquence de sac de mots (BOW) locaux: la séquence vidéo est segmentée temporellement en utilisant la technique de fenêtre glissante et chacun des segments ainsi obtenu est représenté par BOW des points d'intérêt lui appartenant. Le premier niveau de notre système de classification séquentiel hybride consiste à appliquer les séparateurs à vaste marge (SVM) en tant que classifieur de bas niveau afin de convertir les BOWs locaux en des vecteurs de probabilités des classes d'activité. Les séquences de vecteurs de probabilité ainsi obtenues sot utilisées comme l'entrées de classifieur séquentiel conditionnel champ aléatoire caché (HCRF). Ce dernier permet de classifier d'une manière discriminante les séries temporelles tout en modélisant leurs structures internes via les états cachés. Nous avons évalué notre approche sur des bases publiques ayant des caractéristiques diverses. Les résultats atteints semblent être intéressant par rapport à celles des travaux de l'état de l'art. De plus, nous avons montré que l'utilisation de classifieur de bas niveau permet d'améliorer la performance de système de reconnaissance vue que le classifieur séquentiel HCRF traite directement des informations sémantiques des BOWs locaux, à savoir la probabilité de chacune des activités relativement au segment en question. De plus, les vecteurs de probabilités ont une dimension faible ce qui contribue à éviter le problème de sur apprentissage qui peut intervenir si la dimension de vecteur de caractéristique est plus importante que le nombre des données; ce qui le cas lorsqu'on utilise les BOWs qui sont généralement de dimension élevée. L'estimation les paramètres du HCRF dans un espace de dimension réduite permet aussi de réduire le temps d'entrainement / Human activity recognition (HAR) from video sequences is one of the major active research areas of computer vision. There are numerous application HAR systems, including video-surveillance, search and automatic indexing of videos, and the assistance of frail elderly. This task remains a challenge because of the huge variations in the way of performing activities, in the appearance of the person and in the variation of the acquisition conditions. The main objective of this thesis is to develop an efficient HAR method that is robust to different sources of variability. Approaches based on interest points have shown excellent state-of-the-art performance over the past years. They are generally related to global classification methods as these primitives are temporally and spatially disordered. More recent studies have achieved a high performance by modeling the spatial and temporal context of interest points by encoding, for instance, the neighborhood of the interest points over several scales. In this thesis, we propose a method of activity recognition based on a hybrid model Support Vector Machine - Hidden Conditional Random Field (SVM-HCRF) that models the sequential aspect of activities while exploiting the robustness of interest points in real conditions. We first extract the interest points and show their robustness with respect to the person's identity by a multilinear tensor analysis. These primitives are then represented as a sequence of local "Bags of Words" (BOW): The video is temporally fragmented using the sliding window technique and each of the segments thus obtained is represented by the BOW of interest points belonging to it. The first layer of our hybrid sequential classification system is a Support Vector Machine that converts each local BOW extracted from the video sequence into a vector of activity classes’ probabilities. The sequence of probability vectors thus obtained is used as input of the HCRF. The latter permits a discriminative classification of time series while modeling their internal structures via the hidden states. We have evaluated our approach on various human activity datasets. The results achieved are competitive with those of the current state of art. We have demonstrated, in fact, that the use of a low-level classifier (SVM) improves the performance of the recognition system since the sequential classifier HCRF directly exploits the semantic information from local BOWs, namely the probability of each activity relatively to the current local segment, rather than mere raw information from interest points. Furthermore, the probability vectors have a low-dimension which prevents significantly the risk of overfitting that can occur if the feature vector dimension is relatively high with respect to the training data size; this is precisely the case when using BOWs that generally have a very high dimension. The estimation of the HCRF parameters in a low dimension allows also to significantly reduce the duration of the HCRF training phase

Artificial intelligence tools, which are fast, robust and adaptive can overcome the drawbacks of traditional solutions for several power systems problems. In this work, applications of AI techniques have been studied for solving two important problems in power systems. The first problem is static security evaluation (SSE). The objective of SSE is to identify the contingencies in planning and operations of power systems. Numerical conventional solutions are time-consuming, computationally expensive, and are not suitable for online applications. SSE may be considered as a binary-classification, multi-classification or regression problem. In this work, multi-support vector machine is combined with several evolutionary computation algorithms, including particle swarm optimization (PSO), differential evolution, Ant colony optimization for the continuous domain, and harmony search techniques to solve the SSE. Moreover, support vector regression is combined with modified PSO with a proposed modification on the inertia weight in order to solve the SSE. Also, the correct accuracy of classification, the speed of training, and the final cost of using power equipment heavily depend on the selected input features. In this dissertation, multi-object PSO has been used to solve this problem. Furthermore, a multi-classifier voting scheme is proposed to get the final test output. The classifiers participating in the voting scheme include multi-SVM with different types of kernels and random forests with an adaptive number of trees. In short, the development and performance of different machine learning tools combined with evolutionary computation techniques have been studied to solve the online SSE. The performance of the proposed techniques is tested on several benchmark systems, namely the IEEE 9-bus, 14-bus, 39-bus, 57-bus, 118-bus, and 300-bus power systems. The second problem is the non-convex, nonlinear, and non-differentiable economic dispatch (ED) problem. The purpose of solving the ED is to improve the cost-effectiveness of power generation. To solve ED with multi-fuel options, prohibited operating zones, valve point effect, and transmission line losses, genetic algorithm (GA) variant-based methods, such as breeder GA, fast navigating GA, twin removal GA, kite GA, and United GA are used. The IEEE systems with 6-units, 10-units, and 15-units are used to study the efficiency of the algorithms.

Electric power systems are among the most complex man made systems on the world. Most of the time, they operate under quasi-steady state. With the ever increasing load demand and the advent of the deregulated power market recently, the power systems are pushed more often to operate close to their design limits and with more uncertainty of the system operating mode. With the increasing complexity and more interconnected systems, power systems are operating closer to their performance limits. As a result, maintaining system security and facilitating efficient system operation have been challenging tasks. Transmission systems are considered the most vital components in power systems connecting both generating/substation and consumer areas with several interconnected networks. In the past, they were owned by regulated, vertically integrated utility companies. They have been designed and operated so that conditions in close proximity to security boundaries are not frequently encountered. However, in the new open access environment, operating conditions tend to be much closer to security boundaries, as transmission use is increasing in sudden and unpredictable directions. Transmission unbundling, coupled with other regulatory requirements, has made new transmission facility construction more difficult. Unfortunately these transmission lines are frequently subjected to a wide variety of faults. Thus, providing proper protective functions for them is essential. Generally the protection of Extra High Voltage (EHV) and Ultra High Voltage (UHV) transmission lines are carried out by the use of distance relays in view of the fact that they provide fast fault clearance and system coordination. Transmission line relaying involves detection, classification and location of transmission line faults. Fast detections of faults enable quick isolation of the faulty line from service and hence, protecting it from the harmful effects of fault. Classification of faults means identification of the type of fault and faulted line section, and this information is required for finding the fault location and assessing the extent of repair work to be carried out. Accurate fault location is necessary for facilitating quick repair and restoration of the line, to improve the reliability and availability of the power supply. Generally, the protection system using conventional distance relaying algorithm involves three zones. The first zone (Z1) of the relay is set to detect faults on 80%90% of the protected line without any intentional time delay. The second zone (Z2) is set to protect the remainder of the line plus an adequate margin. Second zone relays are time delayed for 1530 cycles to coordinate with relays at remote bus. The settings of the third zone (Z3) ideally will cover the protected line, plus all of the longest line leaving the remote station. Z3 of a distance relay is used to provide the remote backup protection in case of the failure of the primary protection. Since Z3 covers an adjacent line, a large infeed (outfeed) from the remote terminal causes the relay to underreach (overreach). Thus, a very large load at the remote terminal may cause distance relays to mal-operate. Settings for conventional distance relays are selected to avoid overreach/underreach operation under the worst case scenarios. Studies of significant power system disturbances reported by North American Electric Reliability Council (NERC) indicate that protective relays are involved, one way or another, in 75% of the major disturbances and the most troublesome ones are backup protection relays. With their limited view of the interconnected network based on their locally measured inputs, conventional backup protection relays generally take actions to protect a localized region of the network without considering the impact on the whole network. Relay mal-operations or unintended operations due to overload, power swing, and relay hidden failure are the main factors contributing to the blackouts. Most of the problems are associated with relays tripping too many healthy lines. Since a relay makes the decision automatically to remove a component from the system according to its internal mechanism, the relay mal-operation or unintended operation can make an effective influence on the system stability. Approaches to reduce the relay misbehavior need to be identified. Real time monitoring tools to assess the relay misbehavior are needed, providing the system operator, the accurate information about unfolding events. Existing transmission line protection scheme still has drawbacks. Advanced fault analysis mechanism to enhance the system dependability and security simultaneously is desirable. Relay settings play a significant role in major blackouts. So correct settings should be calculated and coordinated by suitable studies. Attempts are to be made to employ highly accurate AI techniques in protective system implementation. The research work focussed on developing knowledge based intelligent tools for the improving the transmission system security. A process to obtain knowledgebase using SVMs for ready post-fault diagnosis purpose is developed. SVMs are used as Intelligent tool for identifying the faulted line that is emanating from a substation and finding the distance from the substation. The approach uses phasor values of the line voltages and currents after the fault has been detected. The approach is particularly important for post-fault diagnosis of any mal-operation of relays following a disturbance in the neighboring line connected to the same substation. This may help in improving the fault monitoring/diagnosis process and coordination of the protective relays, thus assuring secure operation of the power systems. The approach based on SVMs, exploits the first part of this goal. For comparison, a classifier and regression tools based on the RBFNNs was also investigated. The RBFNNs and SVM networks are introduced and considered as an appropriate tool for pattern recognition problems. Results on a practical 24Bus equivalent EHV transmission system of Indian Southern region and on IEEE39 bus New England system are presented to illustrate the proposed method. In a large connected power network, the number of generators are more in number and their set patterns number will be large. As the line flows are sensitive to generator set patterns, it is difficult to consider all the combinations of generators while simulating the training and testing patterns as input to SVMs. To simulate the training and testing patterns corresponding to possible changes in line flows to meet the load in the present deregulated environment, line flow sensitive generators set to be identified/merit-listed. In this regard, to identify the most sensitive generators for a particular line of interest, a method from the literature is adopted and developed a software program based on the graph theory concepts. Case studies on generator contributions towards loads and transmission lines are illustrated on an equivalent 33bus system, a part of Indian Northern grid with major part of Uttar Pradesh and also with an equivalent 246bus system of practical Indian Southern grid. A distance relay coordination approach is proposed using detailed simulation studies, taking into account various operating conditions and fault resistances. Support Vector Machines as a pattern classifier is used for obtaining distance relay coordination. The scheme uses the apparent impedance values observed during fault as inputs. SVMs are used to build the underlying concept between reach of different zones and the impedance trajectory during fault. An improved performance with the use of SVMs, keeping the reach when faced with different fault conditions as well as line flow changes are illustrated with an equivalent 246bus system of practical Indian Southern grid and also with an equivalent 265bus system of practical Indian Western grid. A strategy of Supportive System is described to aid the conventional protection philosophy in combating situations where protection systems are mal-operated and/or information is missing and provide selective and secure coordination. Highly accurate identification/discrimination of zones plays a key role in effective implementation of the region-wide supportive system. This typically requires a multiclass SVM classifier to effectively analyze/build the underlying concept between reach of different zones and the apparent impedance trajectory during fault. Different multiclass methods are compared for their performance with respect to accuracy, number of iterations, number of support vectors, training and testing time. The performance analysis of these methods is presented on three data sets belonging to the training and testing patterns of three supportive systems for a region, part of a network, which is an equivalent 265bus system of practical Indian Western grid.

The growth of multimedia content in recent years motivated the research on audio classification and content retrieval area. In this thesis, a general environmental audio classification and retrieval approach is proposed in which higher level semantic classes (outdoor, nature, meeting and violence) are obtained from lower level acoustic classes (emergency alarm, car horn, gun-shot, explosion, automobile, motorcycle, helicopter, wind, water, rain, applause, crowd and laughter). In order to classify an audio sample into acoustic classes, MPEG-7 audio features, Mel Frequency Cepstral Coefficients (MFCC) feature and Zero Crossing Rate (ZCR) feature are used with Hidden Markov Model (HMM) and Support Vector Machine (SVM) classifiers. Additionally, a new classification method is proposed using Genetic Algorithm (GA) for classification of semantic classes. Query by Example (QBE) and keyword-based query capabilities are implemented for content retrieval.

The main contribution of this thesis is the concept of Fenchel duality with a focus on its application in the field of machine learning problems and image restoration tasks. We formulate a general optimization problem for modeling support vector machine tasks and assign a Fenchel dual problem to it, prove weak and strong duality statements as well as necessary and sufficient optimality conditions for that primal-dual pair. In addition, several special instances of the general optimization problem are derived for different choices of loss functions for both the regression and the classifification task. The convenience of these approaches is demonstrated by numerically solving several problems. We formulate a general nonsmooth optimization problem and assign a Fenchel dual problem to it. It is shown that the optimal objective values of the primal and the dual one coincide and that the primal problem has an optimal solution under certain assumptions. The dual problem turns out to be nonsmooth in general and therefore a regularization is performed twice to obtain an approximate dual problem that can be solved efficiently via a fast gradient algorithm. We show how an approximate optimal and feasible primal solution can be constructed by means of some sequences of proximal points closely related to the dual iterates. Furthermore, we show that the solution will indeed converge to the optimal solution of the primal for arbitrarily small accuracy. Finally, the support vector regression task is obtained to arise as a particular case of the general optimization problem and the theory is specialized to this problem. We calculate several proximal points occurring when using difffferent loss functions as well as for some regularization problems applied in image restoration tasks. Numerical experiments illustrate the applicability of our approach for these types of problems.

The generalization abilities of machine learning algorithms often depend on the algorithms’ initialization, parameter settings, training sets, or feature selections. For instance, SVM classifier performance largely relies on whether the selected kernel functions are suitable for real application data. To enhance the performance of individual classifiers, this dissertation proposes classifier fusion models using computational intelligence knowledge to combine different classifiers. The first fusion model called T1FFSVM combines multiple SVM classifiers through constructing a fuzzy logic system. T1FFSVM can be improved by tuning the fuzzy membership functions of linguistic variables using genetic algorithms. The improved model is called GFFSVM. To better handle uncertainties existing in fuzzy MFs and in classification data, T1FFSVM can also be improved by applying type-2 fuzzy logic to construct a type-2 fuzzy classifier fusion model (T2FFSVM). T1FFSVM, GFFSVM, and T2FFSVM use accuracy as a classifier performance measure. AUC (the area under an ROC curve) is proved to be a better classifier performance metric. As a comparison study, AUC-based classifier fusion models are also proposed in the dissertation. The experiments on biomedical datasets demonstrate promising performance of the proposed classifier fusion models comparing with the individual composing classifiers. The proposed classifier fusion models also demonstrate better performance than many existing classifier fusion methods. The dissertation also studies one interesting phenomena in biology domain using machine learning and classifier fusion methods. That is, how protein structures and sequences are related each other. The experiments show that protein segments with similar structures also share similar sequences, which add new insights into the existing knowledge on the relation between protein sequences and structures: similar sequences share high structure similarity, but similar structures may not share high sequence similarity.

El propòsit d'aquesta tesi és examinar les aportacions a l'estudi de la predicció de la taxa de canvi basada en l'ús de tècniques d'aprenentatge automàtic. Aquestes aportacions es veuen facilitades i millorades per l'ús de variables econòmiques, indicadors tècnics i variables de tipus ‘business and consumer survey’. Aquesta investigació s’organitza entorn d’una recopilació de quatre articles. L'objectiu de cadascun dels quatre treballs de recerca d'aquesta tesi és el de contribuir a l'avanç del coneixement sobre els efectes i mecanismes mitjançant els quals l'ús de variables econòmiques, indicadors tècnics, variables de tipus ‘business and consumer survey’, i la selecció dels paràmetres de models predictius són capaços de millorar les prediccions de la taxa de canvi. Fent ús d'una tècnica de predicció no lineal, el primer article d'aquesta tesi es centra majoritàriament en l'impacte que tenen l'ús de variables econòmiques i la selecció dels paràmetres dels models en les prediccions de la taxa de canvi per a dos països. L'últim experiment d'aquest primer article fa ús de la taxa de canvi del període anterior i d'indicadors econòmics com a variables d'entrada en els models predictius. El segon article d'aquesta tesi analitza com la combinació de mitjanes mòbils, variables de tipus ‘business and consumer survey’ i la selecció dels paràmetres dels models milloren les prediccions del canvi per a dos països. A diferència del primer article, aquest segon treball de recerca afegeix mitjanes mòbils i variables de tipus ‘business and consumer survey’ com a variables d'entrada en els models predictius, i descarta l'ús de variables econòmiques. Un dels objectius d'aquest segon article és determinar el possible impacte de les variables de tipus ‘business and consumer survey’ en les taxes de canvi. El tercer article d'aquesta tesi té els mateixos objectius que el segon, però amb l'excepció que l'anàlisi abasta les taxes de canvi de set països. El quart article de la tesi compta amb els mateixos objectius que l'article anterior, però amb la diferència que fa ús d'un sol indicador tècnic. En general, l'enfocament d'aquesta tesi pretén examinar diferents alternatives per a millorar les prediccions del tipus de canvi a través de l'ús de màquines de suport vectorial. Una combinació de variables i la selecció dels paràmetres dels models predictius ajudaran a aconseguir aquest propòsit. / El propósito de esta tesis es examinar las aportaciones al estudio de la predicción de la tasa de cambio basada en el uso de técnicas de aprendizaje automático. Dichas aportaciones se ven facilitadas y mejoradas por el uso de variables económicas, indicadores técnicos y variables de tipo ‘business and consumer survey’. Esta investigación está organizada en un compendio de cuatro artículos. El objetivo de cada uno de los cuatro trabajos de investigación de esta tesis es el de contribuir al avance del conocimiento sobre los efectos y mecanismos mediante los cuales el uso de variables económicas, indicadores técnicos, variables de tipo ‘business and consumer survey’, y la selección de los parámetros de modelos predictivos son capaces de mejorar las predicciones de la tasa de cambio. Haciendo uso de una técnica de predicción no lineal, el primer artículo de esta tesis se centra mayoritariamente en el impacto que tienen el uso de variables económicas y la selección de los parámetros de los modelos en las predicciones de la tasa de cambio para dos países. El último experimento de este primer artículo hace uso de la tasa de cambio del periodo anterior y de indicadores económicos como variables de entrada en los modelos predictivos. El segundo artículo de esta tesis analiza cómo la combinación de medias móviles, variables de tipo ‘business and consumer survey’ y la selección de los parámetros de los modelos mejoran las predicciones del cambio para dos países. A diferencia del primer artículo, este segundo trabajo de investigación añade medias móviles y variables de tipo ‘business and consumer survey’ como variables de entrada en los modelos predictivos, y descarta el uso de variables económicas. Uno de los objetivos de este segundo artículo es determinar el posible impacto de las variables de tipo ‘business and consumer survey’ en las tasas de cambio. El tercer artículo de esta tesis tiene los mismos objetivos que el segundo, pero con la salvedad de que el análisis abarca las tasas de cambio de siete países. El cuarto artículo de esta tesis cuenta con los mismos objetivos que el artículo anterior, pero con la diferencia de que hace uso de un solo indicador técnico. En general, el enfoque de esta tesis pretende examinar diferentes alternativas para mejorar las predicciones del tipo de cambio a través del uso de máquinas de soporte vectorial. Una combinación de variables y la selección de los parámetros de los modelos predictivos ayudarán a conseguir este propósito. / The purpose of this thesis is to examine the contribution made by machine learning techniques on exchange rate forecasting. Such contributions are facilitated and enhanced by the use of fundamental economic variables, technical indicators and business and consumer survey variables as inputs in the forecasting models selected. This research has been organized in a compendium of four articles. The aim of each of these four articles is to contribute to advance our knowledge on the effects and means by which the use of fundamental economic variables, technical indicators, business and consumer surveys, and a model’s free-parameters selection is capable of improving exchange rate predictions. Through the use of a non-linear forecasting technique, one research paper examines the effect of fundamental economic variables and a model’s parameters selection on exchange rate forecasts, whereas the other three articles concentrate on the effect of technical indicators, a model’s parameters selection and business and consumer surveys variables on exchange rate forecasting. The first paper of this thesis has the objective of examining fundamental economic variables and a forecasting model’s parameters in an effort to understand the possible advantages or disadvantages these variables may bring to the exchange rate predictions in terms of forecasting performance and accuracy. The second paper of this thesis analyses how the combination of moving averages, business and consumer surveys and a forecasting model’s parameters improves exchange rate predictions. Compared to the first paper, this second paper adds moving averages and business and consumer surveys variables as inputs to the forecasting model, and disregards the use of fundamental economic variables. One of the goals of this paper is to determine the possible effects of business and consumer surveys on exchange rates. The third paper of this thesis has the same objectives as the second paper, but its analysis is expanded by taking into account the exchange rates of 7 countries. The fourth paper in this thesis takes a similar approach as the second and third papers, but makes use of a single technical indicator. In general, this thesis focuses on the improvement of exchange rate predictions through the use of support vector machines. A combination of variables and a model’s parameters selection enhances the way to achieve this purpose.

Content Based Image Retrieval (CBIR) is an important research area in the field of multimedia information retrieval. The application of CBIR in the medical domain has been attempted before, however the use of CBIR in medical diagnostics is a daunting task. The goal of diagnostic medical image retrieval is to provide diagnostic support by displaying relevant past cases, along with proven pathologies as ground truths. Moreover, medical image retrieval can be extremely useful as a training tool for medical students and residents, follow-up studies, and for research purposes. Despite the presence of an impressive amount of research in the area of CBIR, its acceptance for mainstream and practical applications is quite limited. The research in CBIR has mostly been conducted as an academic pursuit, rather than for providing the solution to a need. For example, many researchers proposed CBIR systems where the image database consists of images belonging to a heterogeneous mixture of man-made objects and natural scenes while ignoring the practical uses of such systems. Furthermore, the intended use of CBIR systems is important in addressing the problem of "Semantic Gap". Indeed, the requirements for the semantics in an image retrieval system for pathological applications are quite different from those intended for training and education. Moreover, many researchers have underestimated the level of accuracy required for a useful and practical image retrieval system. The human eye is extremely dexterous and efficient in visual information processing; consequently, CBIR systems should be highly precise in image retrieval so as to be useful to human users. Unsurprisingly, due to these and other reasons, most of the proposed systems have not found useful real world applications. In this dissertation, an attempt is made to address the challenging problem of developing a retrieval system for medical diagnostics applications. More specifically, a system for semantic retrieval of Magnetic Resonance (MR) images in 3D brain volumes is proposed. The proposed retrieval system has a potential to be useful for clinical experts where the human eye may fail. Previously proposed systems used imprecise segmentation and feature extraction techniques, which are not suitable for precise matching requirements of the image retrieval in this application domain. This dissertation uses multiscale representation for image retrieval, which is robust against noise and MR inhomogeneity. In order to achieve a higher degree of accuracy in the presence of misalignments, an image registration based retrieval framework is developed. Additionally, to speed-up the retrieval system, a fast discrete wavelet based feature space is proposed. Further improvement in speed is achieved by semantically classifying of the human brain into various "Semantic Regions", using an SVM based machine learning approach. A novel and fast identification system is proposed for identifying a 3D volume given a 2D image slice. To this end, we used SVM output probabilities for ranking and identification of patient volumes. The proposed retrieval systems are tested not only for noise conditions but also for healthy and abnormal cases, resulting in promising retrieval performance with respect to multi-modality, accuracy, speed and robustness. This dissertation furnishes medical practitioners with a valuable set of tools for semantic retrieval of 2D images, where the human eye may fail. Specifically, the proposed retrieval algorithms provide medical practitioners with the ability to retrieve 2D MR brain images accurately and monitor the disease progression in various lobes of the human brain, with the capability to monitor the disease progression in multiple patients simultaneously. Additionally, the proposed semantic classification scheme can be extremely useful for semantic based categorization, clustering and annotation of images in MR brain databases. This research framework may evolve in a natural progression towards developing more powerful and robust retrieval systems. It also provides a foundation to researchers in semantic based retrieval systems on how to expand existing toolsets for solving retrieval problems.