Multi-target tracking is the process that allows the extraction of object motion patterns of interest from a scene. Motion patterns are often described through metadata representing object locations and shape information. In the first part of this thesis we discuss the state-of-the-art methods aimed at accomplishing this task on monocular views and also analyse the methods for evaluating their performance. The second part of the thesis describes our research contribution to these topics. We begin presenting a method for multi-target tracking based on track-before-detect (MTTBD) formulated as a particle filter. The novelty involves the inclusion of the target identity (ID) into the particle state, which enables the algorithm to deal with an unknown and unlimited number of targets. We propose a probabilistic model of particle birth and death based on Markov Random Fields. This model allows us to overcome the problem of the mixing of IDs of close targets. We then propose three evaluation measures that take into account target-size variations, combine accuracy and cardinality errors, quantify long-term tracking accuracy at different accuracy levels, and evaluate ID changes relative to the duration of the track in which they occur. This set of measures does not require pre-setting of parameters and allows one to holistically evaluate tracking performance in an application-independent manner. Lastly, we present a framework for multi-target localisation applied on scenes with a high density of compact objects. Candidate target locations are initially generated by extracting object features from intensity maps using an iterative method based on a gradient-climbing technique and an isocontour slicing approach. A graph-based data association method for multi-target tracking is then applied to link valid candidate target locations over time and to discard those which are spurious. This method can deal with point targets having indistinguishable appearance and unpredictable motion. MT-TBD is evaluated and compared with state-of-the-art methods on real-world surveillance.

Consider the graph induced by Z^d, equipped with uniformly elliptic random conductances on the edges. At time 0, place a Poisson point process of particles on Z^d and let them perform independent simple random walks with jump probabilities proportional to the conductances. It is well known that without conductances (i.e., all conductances equal to 1), an infection started from the origin and transmitted between particles that share a site spreads in all directions with positive speed. We show that a local mixing result holds for random conductance graphs and prove the existence of a special percolation structure called the Lipschitz surface. Using this structure, we show that in the setup of particles on a uniformly elliptic graph, an infection also spreads with positive speed in any direction. We prove the robustness of the framework by extending the result to infection with recovery, where we show positive speed and that the infection survives indefinitely with positive probability.

The goal of this study is to try to understand multi-strain diseases with the presence of cross-immunity by using mathematical models and other mathematical tools. Cross-immunity occurs when a host who is exposed to one disease, or one strain of a disease, develops resistance or partial resistance to related diseases or strains. It is an important factor in the epidemiology of diseases prone to mutation. This work includes modelling influenza in both presence and absence of controls. It also includes modelling malaria when cross-species immunity is present. In addition, vector-bias of mosquitoes to infected humans is also studied in the single-strain malaria model.

Les futurs systèmes de communication sans fil devront utiliser des architectures cellulaires hétérogènes composées de grandes cellules (macro) plus performantes et de petites cellules (femto, micro, ou pico) très denses, afin de soutenir la demande de débit en augmentation exponentielle au niveau de la couche physique. Ces structures provoquent un niveau d'interférence sans précèdent à l'intérieur, comme à l'extérieur des cellules, qui doit être atténué ou, idéalement, exploité afin d'améliorer l'efficacité spectrale globale du réseau. Des techniques comme le MIMO à grande échelle (dit massive MIMO), la coopération, etc., qui contribuent aussi à la gestion des interférences, vont encore augmenter la taille des grandes architectures hétérogènes, qui échappent ainsi à toute possibilité d'analyse théorique par des techniques statistiques traditionnelles.Par conséquent, dans cette thèse, nous allons appliquer et améliorer des résultats connus de la théorie des matrices aléatoires à grande échelle (RMT) afin d'analyser le problème d'interférence et de proposer de nouveaux systèmes de précodage qui s'appuient sur les résultats acquis par l'analyse du système à grande échelle. Nous allons d'abord proposer et analyser une nouvelle famille de précodeurs qui réduit la complexité de calcul de précodage pour les stations de base équipées d'un grand nombre d'antennes, tout en conservant la plupart des capacités d'atténuation d'interférence de l'approche classique et le caractère quasi-optimal du précodeur regularised zero forcing. Dans un deuxième temps, nous allons proposer une variation de la structure de précodage linéaire optimal (obtenue pour de nombreuses mesures de performance) qui permet de réduire le niveau d'interférence induit aux autres cellules. Ceci permet aux petites cellules d'atténuer efficacement les interférences induites et reçues au moyen d'une coopération minimale. Afin de faciliter l'utilisation de l'approche analytique RMT pour les futures générations de chercheurs, nous fournissons également un tutoriel exhaustif sur l'application pratique de la RMT pour les problèmes de communication en début du manuscrit. / Future wireless communication systems will need to feature multi cellular heterogeneous architectures consisting of improved macro cells and very dense small cells, in order to support the exponentially rising demand for physical layer throughput. Such structures cause unprecedented levels of inter and intra cell interference, which needs to be mitigated or, ideally, exploited in order to improve overall spectral efficiency of the communication network. Techniques like massive multiple input multiple output (MIMO), cooperation, etc., that also help with interference management, will increase the size of the already large heterogeneous architectures to truly enormous networks, that defy theoretical analysis via traditional statistical methods.Accordingly, in this thesis we will apply and improve the already known framework of large random matrix theory (RMT) to analyse the interference problem and propose solutions centred around new precoding schemes, which rely on large system analysis based insights. First, we will propose and analyse a new family of precoding schemes that reduce the computational precoding complexity of base stations equipped with a large number of antennas, while maintaining most of the interference mitigation capabilities of conventional close-to-optimal regularized zero forcing. Second, we will propose an interference aware linear precoder, based on an intuitive trade-off and recent results on multi cell regularized zero forcing, that allows small cells to effectively mitigate induced interference with minimal cooperation. In order to facilitate utilization of the analytic RMT approach for future generations of interested researchers, we will also provide a comprehensive tutorial on the practical application of RMT in communication problems.

The purpose of this study is to examine the differences in the extent of customer e-loyalty between the two fashion e-commerce formats single-brand and multi-brand e-retailer. A quantitative research was conducted by means of an electronic questionnaire examining different factors influencing e-loyalty as well as e-loyalty directly. The findings showed similar values for both fashion e-retail formats with slightly higher numbers for single-brand e-retailer. This finding was surprising since it was assumed that single-brand e-retailer would rather attract loyal customers than multi-brand e-retailers. In the case of both formats, however, there is space for improvement to increase customer loyalty. / Program: Master Programme in Fashion Management

Dans cette thèse nous développons de nouvelles techniques pour résoudre les problèmes d'optimisation multi-critère. Ces problèmes se posent naturellement dans de nombreux domaines d'application (sinon tous) où les choix sont évalués selon différents critères conflictuels (coûts et performance par exemple). Contrairement au cas de l'optimisation classique, de tels problèmes n'admettent pas en général un optimum unique mais un ensemble de solutions incomparables, aussi connu comme le front de Pareto, qui représente les meilleurs compromis possibles entre les objectifs conflictuels. La contribution majeure de la thèse est le développement d'algorithmes pour trouver ou approximer ces solutions de Pareto pour les problèmes combinatoires difficiles. Plusieurs problèmes de ce type se posent naturellement lors du processus de placement et d'ordonnancement d'une application logicielle sur une architecture multi-coeur comme P2012, qui est actuellement développé par STMicroelectronics. / In this thesis we develop new techniques for solving multi-criteria optimization problems. Such problems arise naturally in many (if not all) application domains where choices are evaluated according to two or more conflicting criteria such as price vs. performance. Unlike ordinary optimization, such problems typically do not admit a unique optimum but a set of incomparable solutions, also known as the Pareto Front, which represent the best possible trade-offs between the conflicting goals. The major contribution of the thesis is the development of algorithms for finding or approximating these Pareto solutions for hard combinatorial problems that arise naturally in the process of mapping and scheduling application software on multi-core architectures such as P2012 which is currently being developed by ST Microelectronics.

Boston University. University Professors Program Senior theses. / PLEASE NOTE: Boston University Libraries did not receive an Authorization To Manage form for this thesis. It is therefore not openly accessible, though it may be available by request. If you are the author or principal advisor of this work and would like to request open access for it, please contact us at open-help@bu.edu. Thank you. / 2031-01-02

In the health-informatics and bio-medical domains, clinicians produce an enormous amount of data which can be complex and high in dimensionality. This scenario includes visual field data, which are used for managing the second leading cause of blindness in the world: glaucoma. Visual field data are the most common type of data collected to diagnose glaucoma in patients, and usually the data consist of 54 or 76 variables (which are referred to as visual field locations). Due to the large number of variables, the six nerve fiber bundles (6NFB), which is a collection of visual field locations in groups, are the standard clusters used in visual field data to represent the physiological traits of the retina. However, with regard to classification accuracy of the data, this research proposes a technique to find other significant spatial clusters of visual field with higher classification accuracy than the 6NFB. This thesis presents a novel clustering technique, namely, Simultaneous Modelling and Clustering (SMC). SMC performs clustering on data based on classification accuracy using heuristic search techniques. The method searches a collection of significant clusters of visual field locations that indicate visual field loss progression. The aim of this research is two-fold. Firstly, SMC algorithms are developed and tested on data to investigate the effectiveness and efficiency of the method using optimisation and classification methods. Secondly, a significant clustering arrangement of visual field, which highly interrelated visual field locations to represent progression of visual field loss with high classification accuracy, is searched to complement the 6NFB in diagnosis of glaucoma. A new clustering arrangement of visual field locations can be used by medical practitioners together with the 6NFB to complement each other in diagnosis of glaucoma in patients. This research conducts extensive experiment work on both visual field and simulated data to evaluate the proposed method. The results obtained suggest the proposed method appears to be an effective and efficient method in clustering visual field data and 3 improving classification accuracy. The key contributions of this work are the novel model-based clustering of visual field data, effective and efficient algorithms for SMC, practical knowledge of visual field data in the diagnosis of glaucoma and the presentation a generic framework for modelling and clustering which is highly applicable to many other dataset/model combinations.

High gradient acceleration is a core challenge of accelerator physics. Achieving high gradients is made challenging by issues relating to rf breakdown and pulsed surface heating, which are caused by intense surface fields in the accelerating cavities. The excitation of multiple harmonically related modes within a cavity could reduce the onset of these effects. The temperature rise from pulsed surface heating can be reduced by lowering the average magnetic surface field squared and rf breakdown could be avoided by creating an asymmetry between the anode and cathode surface electric fields. This thesis will present several different cavity designs that show a reduction in the temperature rise on the surface of over 10% for second and third harmonic cavity structures or an asymmetry in the surface electric anode and cathode fields of a factor of 2. The harmonic mode could have undesirable consequences for beam stability. A study of the longitudinal beam dynamics is included that will derive the equations governing the longitudinal motion and show that the harmonic mode will have a minor and predictable effect on the rf bucket. The Compact LInear Collider (CLIC) is a major contender for the next generation of lepton linear colliders and is made challenging by high power requirements and distribution throughout the linac. A high current drive beam is decelerated parallel to the main linac in order to create the required rf power, which can overcome some of these issues. This thesis will describe a novel design for a CLIC-like accelerating structure, using collinear acceleration through fundamental mode detuned cavities. The design will accommodate interleaved drive and test bunches, such that the drive bunches are decelerated and the test bunches are accelerated within the confines of the same cavity which can result in high transformer ratios. The analytical theory based on the circuit model will be verified by time domain simulations. A multi-harmonic detuned accelerating structure is introduced that exhibits the properties of pulsed surface heating reduction and can be used for collinear acceleration. Time domain simulations will verify the transformer ratio to within 3% of theoretical predictions and the average magnetic field squared reduction will be within 20% of the value calculated from eigenmode simulations.

This thesis focuses on combining data envelopment analysis (DEA) and multi-objective linear programming (MOLP) methods to set targets by referencing peers' performances and decision-makers' (DMs) preferences. A large number of past papers have proven the importance of a company having a target; however, obtaining a feasible but challenging target has always been a difficult topic for companies. Since DEA was proposed in 1978, it has become one of the most popular performance assessment tools. The performance possibility set and efficient frontier established by DEA provide solid and scientific reference information for managers to evaluate an individual's efficiency. Based on the successful experience of DEA in performance assessment, many scholars have mentioned that DEA can be used to set appropriate targets as well; however, traditional DEA models do not include DMs' preference information that is crucial to a target-setting process. Therefore, several MOLP methods have been introduced to include DMs' preferences in the target-setting process based on the DEA efficient frontier and performance possibility set. The trade-off-based method is one of the most popular interactive methods that have been incorporated with DEA. However, there are several gaps in the current research: (1) the trade-off-based method could take so many interactions that no DMs could finish the interactive process; (2) DMs might find it very difficult to provide the preference information required by MOLP models; and (3) DMs cannot have an intuitive view in terms of the efficient frontier. Regarding the gaps above, this thesis proposes three new trade-off-based interactive target-setting models based on the DEA performance possibility set and efficient frontier to improve DMs' experience when setting targets. The three models can work independently or can be combined during the decision-making process. The piecewise linear model uses a piecewise linear assumption to simulate DMs' real utility function. It gradually narrows down the region that could contain DMs' most-preferred solution (MPS) until it reaches an acceptable range. This model could help DMs who have limited time for interaction but want to have a global view of the entire efficient frontier. This model has also been proven very helpful when DMs are not sensitive to close efficient solutions. The prioritized trade-off model provides a new way for a DM to know about the efficient frontier, which allows the DM to explore the efficient frontier following the preferred direction with a series of trade-off tables and trade-off figures as visual aids. The stepwise trade-off model focuses on situations where the number of objectives (outputs/inputs for the DEA model) is quite large and DMs cannot provide all indifference trade-offs between all the objectives simultaneously. To release the DMs' burden, the stepwise model starts from two objectives and gradually includes new objectives in the decision-making process, with the assumption that the indifference trade-offs between previous objectives are fixed, until all objectives are included. All three models have been validated through numerical examples and case studies of a Chinese state-owned bank to help DMs to explore their MPS in the DEA production possibility set.