The importance of attribute vector ambiguity has been largely overlooked by the machine learning community. A pattern recognition problem can be solved in many ways within the scope of machine learning. Neural Networks, Decision Tree Algorithms such as C4.5, Bayesian Classifiers, and Instance Based Learning are the main algorithms. All listed solutions fail to address ambiguity in the attribute vector. The research reported shows, ignoring this ambiguity leads to problems of classifier scalability and issues with instance collection and aggregation. The Algorithm presented accounts for both ambiguity of the attribute vector and class label thus solving both issues of scalability and instance collection. The research also shows that when applied to sanitized data sets, suitable for traditional instance based learning, the presented algorithm performs equally as well.

No description available.

This thesis presents the development of a hydrologic model in the vector environment. Establishing spatial relationship between flow elements is the key for flow routing techniques. Such a relationship is called hydrologic topology, making each flow element know which other elements are upstream and which are downstream. Based on the hydrologic topology established for the flow elements, tools were developed for flow network navigation, drainage area estimation, flow length calculation and drainage divide determination. To apply the tools, data required might be obtained from different sources, which may lead to certain problems that have to do with wrong flow direction of stream lines and, mismatches in location of stream lines with respect to the corresponding drainage area polygons. Procedures to detect such inconsistencies and to correct them have been developed and are presented here. Data inconsistencies correction and parameter computation methods form the basis for the development of a routing model, which would be referred as hydrologic model. The hydrologic model consists of an overland flow routing module, two options for channel routing and a reservoir routing module. Two case studies have been presented to show the application of the tools developed.

Francisella tularensis is a highly virulent Gram-negative bacterium causing the zoonotic disease tularemia. Arthropod-borne transmission plays an important role in transferring the disease to humans. F. tularensis induces very low amounts of pro-inflammatory cytokines during infection, due to inhibition of immune signaling pathways and an unusual structure of its lipopolysaccharide (LPS). To date, there is no vaccine available that is approved for public use, although an attenuated live vaccine strain (LVS) is commonly used as a model of the more infectious Francisella strains. To produce an effective vaccine it is important to understand the lifecycle of F. tularensis, including the interaction with the arthropod hosts. Drosophila melanogaster is a widely used model organism, which is increasingly being used in host-pathogen interaction studies as the immune pathways in flies are evolutionary conserved to the immune pathways in humans. An important part of the immune defense of D. melanogaster as well as of arthropods in general is the production of antimicrobial peptides. These peptides primarily target the bacterial membrane, inhibiting bacterial proliferation or directly killing the bacteria. The aim of this thesis was to establish D. melanogaster as a model for F. tularensis infection and as a model for arthropod vectors of F. tularensis. Also, to use D. melanogaster to further study the interaction between F. tularensis and arthropod vectors, with specific regard to the host immune signaling and arthropod antimicrobial peptides. F. tularensis LVS infects and kills D. melanogaster in a dose-dependent manner. During an infection, bacteria are found inside fly hemocytes, phagocytic blood cells, similar as in human infections. In mammals genes of the intracellular growth locus (igl) are important for virulence. In this work it is shown that the igl genes are also important for virulence in flies. These results demonstrate that D. melanogaster can be used as a model to study F. tularensis-host interactions. LVS induces a prolonged activation of several immune signaling pathways in the fly, but seem to interfere with the JNK signaling pathway, similarly as in mammals. Overexpression of the JNK pathway in flies has a protective effect on fly survival. Relish mutant flies, essentially lacking a production of antimicrobial peptides, succumb quickly to a F. tularensis infection, however, F. tularensis is relatively resistant to individual D. melanogaster antimicrobial peptides. Overexpressing antimicrobial peptide genes in wildtype flies has a protective effect on F. tularensis infection, suggesting that a combination of several antimicrobial peptides is necessary to control F. tularensis. The production of numerous antimicrobial peptides might be why D. melanogaster survives relatively long after infection. An intact structure of the lipid A and of the Kdo core of Francisella LPS is necessary for resistance to antimicrobial peptides and full virulence in flies. These results are similar to previous studies in mammals. In contrast to studies in mammals, genes affecting the O-antigen of F. tularensis LPS are not necessary for virulence in flies. In conclusion, this thesis work shows that D. melanogaster can be used as a model for studying F. tularensis-host interactions. LVS activates several immune pathways during infection, but interfere with the JNK pathway. Overexpressing the JNK pathway results in increased survival of flies infected with LVS. Despite rather high resistance to individual antimicrobial peptides, exposure to a combination of several D. melanogaster antimicrobial peptides reduces the virulence of F. tularensis.

Classification algorithms have been widely used in many application domains. Most of these domains deal with massive collection of data and hence demand classification algorithms that scale well with the size of the data sets involved. A classification algorithm is said to be scalable if there is no significant increase in time and space requirements for the algorithm (without compromising the generalization performance) when dealing with an increase in the training set size. Support Vector Machine (SVM) is one of the most celebrated kernel based classification methods used in Machine Learning. An SVM capable of handling large scale classification problems will definitely be an ideal candidate in many real world applications. The training process involved in SVM classifier is usually formulated as a Quadratic Programing(QP) problem. The existing solution strategies for this problem have an associated time and space complexity that is (at least) quadratic in the number of training points. This makes the SVM training very expensive even on classification problems having a few thousands of training examples. This thesis addresses the scalability of the training algorithms involved in both two class and multiclass Support Vector Machines. Efficient training schemes reducing the space and time requirements of the SVM training process are proposed as possible solutions. The classification schemes discussed in the thesis for handling large scale two class classification problems are a) Two selective sampling based training schemes for scaling Non-linear SVM and b) Clustering based approaches for handling unbalanced data sets with Core Vector Machine. To handle large scale multicalss classification problems, the thesis proposes Multiclass Core Vector Machine (MCVM), a scalable SVM based multiclass classifier. In MVCM, the multiclass SVM problem is shown to be equivalent to a Minimum Enclosing Ball (MEB) problem and is then solved using a fast approximate MEB finding algorithm. Experimental studies were done with several large real world data sets such as IJCNN1 and Acoustic data sets from LIBSVM page, Extended USPS data set from CVM page and network intrusion detection data sets of DARPA, US Defense used in KDD 99 contest. From the empirical results it is observed that the proposed classification schemes achieve good generalization performance at low time and space requirements. Further, the scalability experiments done with large training data sets have demonstrated that the proposed schemes scale well. A novel soft clustering scheme called Rough Support Vector Clustering (RSVC) employing the idea of Soft Minimum Enclosing Ball Problem (SMEB) is another contribution discussed in this thesis. Experiments done with a synthetic data set and the real world data set namely IRIS, have shown that RSVC finds meaningful soft cluster abstractions.

Detection of targets in hyperspectral images is a specific case of one-class classification. It is particularly relevant in the area of remote sensing and has received considerable interest in the past few years. The thesis proposes the use of wavelet functions as kernels with Support Vector Data Description for target detection in hyperspectral images. Specifically, it proposes the Adaptive Wavelet Kernel Support Vector Data Description (AWK-SVDD) that learns the optimal wavelet function to be used given the target signature. The performance and computational requirements of AWK-SVDD is compared with that of existing methods and other wavelet functions. An introduction to target detection and target detection in the context of hyperspectral images is given. This thesis also includes an overview of the thesis and lists the contributions of the thesis. A brief mathematical background into one-class classification in reference to target detection is included. Also described are the existing methods and introduces essential concepts relevant to the proposed approach. The use of wavelet functions as kernels with Support Vector Data Description, the conditions for use of wavelet functions and the use of two functions in order to form the kernel are checked and analyzed. The proposed approach, AWKSVDD, is mathematically described. The details of the implementation and the results when applied to the Urban dataset of hyperspectral images with a random target signature are given. The results confirm the better performance of AWK-SVDD compared to conventional kernels, wavelet kernels and the two-function Morlet-Radial Basis Function kernel. The problems faced with convergence during the Support Vector Data Description optimization are discussed. The thesis concludes with the suggestions for future work.

Face detection is a preprocessing step for face recognition algorithms. It is the localization of face/faces in an image or image sequence. Once the face(s) are localized, other computer vision algorithms such as face recognition, image compression, camera auto focusing etc are applied. Because of the multiple usage areas, there are many research efforts in face processing. Face detection is a challenging computer vision problem because of lighting conditions, a high degree of variability in size, shape, background, color, etc. To build fully automated systems, robust and efficient face detection algorithms are required. Numerous techniques have been developed to detect faces in a single image; in this project we have used a classification-based face detection method using Gabor filter features. We have designed five frequencies corresponding to eight orientations channels for extracting facial features from local images. The feature vector based on Gabor filter is used as the input of the face/non-face classifier, which is a Support Vector Machine (SVM) on a reduced feature subspace extracted by using principal component analysis (PCA). Experimental results show promising performance especially on single face images where 78% accuracy is achieved with 0 false acceptances.

Stylized Stroke Fonts (SSF) are stroke fonts which allow variable stroke widths and arbitrary stroke ends. In this thesis project we implement SSF by extending concepts of the traditional vector graphics paradigm, giving what we call Vector Graphics Stylized Stroke Fonts (VGSSF). A stroking algorithm for the new stroke model is developed and implemented in the Opera web browser's internal vector graphics drawing toolkit. Both the HTML 5 Canvas JavaScript interface and SVG fonts are extended to support the new stroke model. An editor and renderer for the SVG font format is implemented inside the browser using the extended Canvas interface. Sample glyphs from Latin and Chinese typefaces are converted to the SVG font format to assess the suitability of the stroke representation. The results are excellent for Chinese Ming typeface glyphs, while there are some minor problems with Latin typeface glyphs. Approximations suggest that VGSSF gives a size reduction of the font definition file by at least 50%, with a potential reduction of around 85% for Chinese typefaces. The processing requirements increase by approximately 20-30% due to extra steps required to render each glyph.

Intelligent Transportation System (ITS) is a system that builds a safe, effective and integrated transportation environment based on advanced technologies. Road signs detection and recognition is an important part of ITS, which offer ways to collect the real time traffic data for processing at a central facility.This project is to implement a road sign recognition model based on AI and image analysis technologies, which applies a machine learning method, Support Vector Machines, to recognize road signs. We focus on recognizing seven categories of road sign shapes and five categories of speed limit signs. Two kinds of features, binary image and Zernike moments, are used for representing the data to the SVM for training and test. We compared and analyzed the performances of SVM recognition model using different features and different kernels. Moreover, the performances using different recognition models, SVM and Fuzzy ARTMAP, are observed.

Application and implication of using complex vectors and complex transformations in solutions of Maxwell’s equations is investigated. Complex vectors are used in complex plane waves and help to represent this type of waves geometrically. It is shown that they are also useful in representing inhomogeneous plane waves in plasma, single-negative and double-negative metamaterials. In specific I will investigate the Otto configuration and Kretschmann configuration and I will show that in order to observe the minimum in reflection coefficient it is necessary for the metal to be lossy. We will compare this to the case of plasmon-like resonance when a PEC periodic structure is illuminated by a plane wave. Complex transformations are crucial in deriving Gaussian beam solutions of paraxial Helmholtz equation from spherical wave solution of Helmholtz equation. Vector Gaussian beams also will be discussed shortly.