Learning Preferences with Kernel-Based Methods

Learning of preference relations has recently received significant attention in machine learning community. It is closely related to the classification and regression analysis and can be reduced to these tasks. However, preference learning involves prediction of ordering of the data points rather than prediction of a single numerical value as in case of regression or a class label as in case of classification. Therefore, studying preference relations within a separate framework facilitates not only better theoretical understanding of the problem, but also motivates development of the efficient algorithms for the task. Preference learning has many applications in domains such as information retrieval, bioinformatics, natural language processing, etc. For example, algorithms that learn to rank are frequently used in search engines for ordering documents retrieved by the query. Preference learning methods have been also applied to collaborative filtering problems for predicting individual customer choices from the vast amount of user generated feedback. In this thesis we propose several algorithms for learning preference relations. These algorithms stem from well founded and robust class of regularized least-squares methods and have many attractive computational properties. In order to improve the performance of our methods, we introduce several non-linear kernel functions. Thus, contribution of this thesis is twofold: kernel functions for structured data that are used to take advantage of various non-vectorial data representations and the preference learning algorithms that are suitable for different tasks, namely efficient learning of preference relations, learning with large amount of training data, and semi-supervised preference learning. Proposed kernel-based algorithms and kernels are applied to the parse ranking task in natural language processing, document ranking in information retrieval, and remote homology detection in bioinformatics domain. Training of kernel-based ranking algorithms can be infeasible when the size of the training set is large. This problem is addressed by proposing a preference learning algorithm whose computation complexity scales linearly with the number of training data points. We also introduce sparse approximation of the algorithm that can be efficiently trained with large amount of data. For situations when small amount of labeled data but a large amount of unlabeled data is available, we propose a co-regularized preference learning algorithm. To conclude, the methods presented in this thesis address not only the problem of the efficient training of the algorithms but also fast regularization parameter selection, multiple output prediction, and cross-validation. Furthermore, proposed algorithms lead to notably better performance in many preference learning tasks considered.

Estudo da mobilização de metais e elementos traços em ambientes aquáticos do semi-árido brasileiro aplicando análises de componentes principais

The spatial and temporal retention of metals has been studied in water and sediments of the Gavião River, Anagé and Tremedal Reservoirs, located in the semi-arid region, Bahia - Brazil, in order to identify trends in the fluxes of metals from the sediments to the water column. The determination of metals was made by ICP OES and ET AAS. The application of statistical methods showed that this aquatic system presents suitable conditions to move Cd2+ and Pb2+ from the water column to the sediment.

Retrieval of kinetic rates in reactions with semi batch liquid phase using ill-posed inverse problem theory

A neural network procedure to solve inverse chemical kinetic problems is discussed in this work. Rate constants are calculated from the product concentration of an irreversible consecutive reaction: the hydrogenation of Citral molecule, a process with industrial interest. Simulated and experimental data are considered. Errors in the simulated data, up to 7% in the concentrations, were assumed to investigate the robustness of the inverse procedure. Also, the proposed method is compared with two common methods in nonlinear analysis; the Simplex and Levenberg-Marquardt approaches. In all situations investigated, the neural network approach was numerically stable and robust with respect to deviations in the initial conditions or experimental noises.

Thermal and structural study of guava (Psidium guajava L) powders obtained by two dehydration methods

Two food products (powders) were obtained by hot-air drying or lyophilisation methods on the whole guava fruits. The powders were characterised by sensory and thermal analyses (TGA-DSC), infrared spectroscopy (IR), X-ray diffraction (XRD) and scanning electron microscopy (SEM). Thermal, morphological and structural characterisations showed a similar behaviour for the two solids. TGA-DSC and IR showed the presence of pectin as the main constituent of solids. A semi-crystalline profile was evidenced by XRD, and lamellar/spherical morphologies were observed by SEM. Sensory analyses revealed an aroma highly related to guava. These value-added food products are an alternative to process guava and avoid loss during postharvest handling.

Transdermal formulations containing human sexual steroids: development and validation of methods and in vitro drug release

In vitro release of bioidentical hormones in four different liposomal transdermal emulsions (containing testosterone, progesterone, estradiol, or estradiol and estriol) was assessed. For this purpose, novel high-performance liquid chromatography methods were developed and validated in an eco-friendly manner and used to determine the in vitro release of such products. The methods were suitable for our intended goal, and the emulsions employed were found to be effective as transporting candidates for the efficient release of hormones in the transdermal delivery of human sexual steroids.

A Semi-Selective agar medium to detect the presence of Xanthomonas axonopodis pv. malvacearum in naturally infected cotton seed

A semi-selective agar medium was developed for detection of Xanthomonas axonopodis pv. malvacearum (Xam) in cotton (Gossypium hirsutum) seed. The basic medium was peptone-sucrose-agar (PSA). Criteria for the semi-selective medium were the typical colony characters of Xam and its pathogenicity on cotton. Several systemic fungicides and antibiotics in different concentrations were tested alone or in combination with others. The final composition of the semi-selective agar medium was established after several attempts in order to inhibit most of the fungal and bacterial saprophytes and favour the development of Xam. It contained PSA + cyclohexamide, cephalexin, pencycuron, triadimenol and tolylfluanid. The bacteria were recovered from naturally infected seeds by the direct plating of 2,000 surface disinfected seeds on the semi-selective medium. The recovery of the pathogen from naturally infected leaf tissues and in dilution plating, on semi-selective medium and on nutrient agar, were comparable. Among the three detection methods tested, the semi-selective medium was found to be the most reliable and quantifiable. Degree of severity of angular leaf spot in the field was not always correlated with the level of infection in the seed. This is the first report of a semi-selective agar medium to detect the presence of Xam in naturally infected cotton seed.

Methods and Models for Accelerating Dynamic Simulation of Fluid Power Circuits

The objective of this dissertation is to improve the dynamic simulation of fluid power circuits. A fluid power circuit is a typical way to implement power transmission in mobile working machines, e.g. cranes, excavators etc. Dynamic simulation is an essential tool in developing controllability and energy-efficient solutions for mobile machines. Efficient dynamic simulation is the basic requirement for the real-time simulation. In the real-time simulation of fluid power circuits there exist numerical problems due to the software and methods used for modelling and integration. A simulation model of a fluid power circuit is typically created using differential and algebraic equations. Efficient numerical methods are required since differential equations must be solved in real time. Unfortunately, simulation software packages offer only a limited selection of numerical solvers. Numerical problems cause noise to the results, which in many cases leads the simulation run to fail. Mathematically the fluid power circuit models are stiff systems of ordinary differential equations. Numerical solution of the stiff systems can be improved by two alternative approaches. The first is to develop numerical solvers suitable for solving stiff systems. The second is to decrease the model stiffness itself by introducing models and algorithms that either decrease the highest eigenvalues or neglect them by introducing steady-state solutions of the stiff parts of the models. The thesis proposes novel methods using the latter approach. The study aims to develop practical methods usable in dynamic simulation of fluid power circuits using explicit fixed-step integration algorithms. In this thesis, twomechanisms whichmake the systemstiff are studied. These are the pressure drop approaching zero in the turbulent orifice model and the volume approaching zero in the equation of pressure build-up. These are the critical areas to which alternative methods for modelling and numerical simulation are proposed. Generally, in hydraulic power transmission systems the orifice flow is clearly in the turbulent area. The flow becomes laminar as the pressure drop over the orifice approaches zero only in rare situations. These are e.g. when a valve is closed, or an actuator is driven against an end stopper, or external force makes actuator to switch its direction during operation. This means that in terms of accuracy, the description of laminar flow is not necessary. But, unfortunately, when a purely turbulent description of the orifice is used, numerical problems occur when the pressure drop comes close to zero since the first derivative of flow with respect to the pressure drop approaches infinity when the pressure drop approaches zero. Furthermore, the second derivative becomes discontinuous, which causes numerical noise and an infinitely small integration step when a variable step integrator is used. A numerically efficient model for the orifice flow is proposed using a cubic spline function to describe the flow in the laminar and transition areas. Parameters for the cubic spline function are selected such that its first derivative is equal to the first derivative of the pure turbulent orifice flow model in the boundary condition. In the dynamic simulation of fluid power circuits, a tradeoff exists between accuracy and calculation speed. This investigation is made for the two-regime flow orifice model. Especially inside of many types of valves, as well as between them, there exist very small volumes. The integration of pressures in small fluid volumes causes numerical problems in fluid power circuit simulation. Particularly in realtime simulation, these numerical problems are a great weakness. The system stiffness approaches infinity as the fluid volume approaches zero. If fixed step explicit algorithms for solving ordinary differential equations (ODE) are used, the system stability would easily be lost when integrating pressures in small volumes. To solve the problem caused by small fluid volumes, a pseudo-dynamic solver is proposed. Instead of integration of the pressure in a small volume, the pressure is solved as a steady-state pressure created in a separate cascade loop by numerical integration. The hydraulic capacitance V/Be of the parts of the circuit whose pressures are solved by the pseudo-dynamic method should be orders of magnitude smaller than that of those partswhose pressures are integrated. The key advantage of this novel method is that the numerical problems caused by the small volumes are completely avoided. Also, the method is freely applicable regardless of the integration routine applied. The superiority of both above-mentioned methods is that they are suited for use together with the semi-empirical modelling method which necessarily does not require any geometrical data of the valves and actuators to be modelled. In this modelling method, most of the needed component information can be taken from the manufacturer’s nominal graphs. This thesis introduces the methods and shows several numerical examples to demonstrate how the proposed methods improve the dynamic simulation of various hydraulic circuits.

Comparison of anaerobic threshold determined by visual and mathematical methods in healthy women

Several methods are used to estimate anaerobic threshold (AT) during exercise. The aim of the present study was to compare AT obtained by a graphic visual method for the estimate of ventilatory and metabolic variables (gold standard), to a bi-segmental linear regression mathematical model of Hinkley's algorithm applied to heart rate (HR) and carbon dioxide output (VCO2) data. Thirteen young (24 ± 2.63 years old) and 16 postmenopausal (57 ± 4.79 years old) healthy and sedentary women were submitted to a continuous ergospirometric incremental test on an electromagnetic braking cycloergometer with 10 to 20 W/min increases until physical exhaustion. The ventilatory variables were recorded breath-to-breath and HR was obtained beat-to-beat over real time. Data were analyzed by the nonparametric Friedman test and Spearman correlation test with the level of significance set at 5%. Power output (W), HR (bpm), oxygen uptake (VO2; mL kg-1 min-1), VO2 (mL/min), VCO2 (mL/min), and minute ventilation (VE; L/min) data observed at the AT level were similar for both methods and groups studied (P > 0.05). The VO2 (mL kg-1 min-1) data showed significant correlation (P < 0.05) between the gold standard method and the mathematical model when applied to HR (r s = 0.75) and VCO2 (r s = 0.78) data for the subjects as a whole (N = 29). The proposed mathematical method for the detection of changes in response patterns of VCO2 and HR was adequate and promising for AT detection in young and middle-aged women, representing a semi-automatic, non-invasive and objective AT measurement.

Manual and semi-automatic quantification of in vivo ¹H-MRS data for the classification of human primary brain tumors

In vivo proton magnetic resonance spectroscopy (¹H-MRS) is a technique capable of assessing biochemical content and pathways in normal and pathological tissue. In the brain, ¹H-MRS complements the information given by magnetic resonance images. The main goal of the present study was to assess the accuracy of ¹H-MRS for the classification of brain tumors in a pilot study comparing results obtained by manual and semi-automatic quantification of metabolites. In vivo single-voxel ¹H-MRS was performed in 24 control subjects and 26 patients with brain neoplasms that included meningiomas, high-grade neuroglial tumors and pilocytic astrocytomas. Seven metabolite groups (lactate, lipids, N-acetyl-aspartate, glutamate and glutamine group, total creatine, total choline, myo-inositol) were evaluated in all spectra by two methods: a manual one consisting of integration of manually defined peak areas, and the advanced method for accurate, robust and efficient spectral fitting (AMARES), a semi-automatic quantification method implemented in the jMRUI software. Statistical methods included discriminant analysis and the leave-one-out cross-validation method. Both manual and semi-automatic analyses detected differences in metabolite content between tumor groups and controls (P < 0.005). The classification accuracy obtained with the manual method was 75% for high-grade neuroglial tumors, 55% for meningiomas and 56% for pilocytic astrocytomas, while for the semi-automatic method it was 78, 70, and 98%, respectively. Both methods classified all control subjects correctly. The study demonstrated that ¹H-MRS accurately differentiated normal from tumoral brain tissue and confirmed the superiority of the semi-automatic quantification method.

Intelligence acquisition methods in cyber domain : examining the circumstantial applicability of cyber intelligence acquisition methods using a hierarchical model

Phenomena in cyber domain, especially threats to security and privacy, have proven an increasingly heated topic addressed by different writers and scholars at an increasing pace – both nationally and internationally. However little public research has been done on the subject of cyber intelligence. The main research question of the thesis was: To what extent is the applicability of cyber intelligence acquisition methods circumstantial? The study was conducted in sequential a manner, starting with defining the concept of intelligence in cyber domain and identifying its key attributes, followed by identifying the range of intelligence methods in cyber domain, criteria influencing their applicability, and types of operatives utilizing cyber intelligence. The methods and criteria were refined into a hierarchical model. The existing conceptions of cyber intelligence were mapped through an extensive literature study on a wide variety of sources. The established understanding was further developed through 15 semi-structured interviews with experts of different backgrounds, whose wide range of points of view proved to substantially enhance the perspective on the subject. Four of the interviewed experts participated in a relatively extensive survey based on the constructed hierarchical model on cyber intelligence that was formulated in to an AHP hierarchy and executed in the Expert Choice Comparion online application. It was concluded that Intelligence in cyber domain is an endorsing, cross-cutting intelligence discipline that adds value to all aspects of conventional intelligence and furthermore that it bears a substantial amount of characteristic traits – both advantageous and disadvantageous – and furthermore that the applicability of cyber intelligence methods is partly circumstantially limited.

Semi-Parametric Weak Instrument Regressions with an Application to the Risk-Return Trade-off

Recent work shows that a low correlation between the instruments and the included variables leads to serious inference problems. We extend the local-to-zero analysis of models with weak instruments to models with estimated instruments and regressors and with higher-order dependence between instruments and disturbances. This makes this framework applicable to linear models with expectation variables that are estimated non-parametrically. Two examples of such models are the risk-return trade-off in finance and the impact of inflation uncertainty on real economic activity. Results show that inference based on Lagrange Multiplier (LM) tests is more robust to weak instruments than Wald-based inference. Using LM confidence intervals leads us to conclude that no statistically significant risk premium is present in returns on the S&P 500 index, excess holding yields between 6-month and 3-month Treasury bills, or in yen-dollar spot returns.

A Semi-Parametric Factor Model for Interest Rates.

Understanding the dynamics of interest rates and the term structure has important implications for issues as diverse as real economic activity, monetary policy, pricing of interest rate derivative securities and public debt financing. Our paper follows a longstanding tradition of using factor models of interest rates but proposes a semi-parametric procedure to model interest rates.

Les croyances des enseignants et des apprenants adultes quant à la rétroaction corrective à l’oral et la pratique réelle en classe de français langue étrangère en Égypte

Différentes études ont montré que le niveau des futurs enseignants, issus des écoles publiques, en français langue étrangère (FLE) en Égypte est assez faible. Ceux-ci font un grand nombre d’erreurs répétitives à l’oral. Quoique ce manque de précision langagière puisse être dû à plusieurs facteurs, il appert que la rétroaction soit une des variables contribuant à ce phénomène (comme le nombre d’étudiants en classe, la durée du cours, l’âge et la motivation des étudiants, les méthodes d’enseignement, etc.). La rétroaction corrective est généralement définie comme toute correction explicite ou implicite de la part de l’enseignant indiquant que la production de l’apprenant est erronée. Elle est considérée comme indispensable dans les classes de langues secondes (LS) (Shmidt, 1983, 2001 ; Long, 1991, 1996 ; Lightbown, 1998). Pour ces raisons, cette étude porte sur la rétroaction corrective et, plus spécifiquement, sur les croyances des enseignants et des apprenants quant à celle-ci, ainsi qu’à son utilisation dans les classes de FLE en Égypte. Les recherches antérieures indiquent que les croyances des enseignants quant à l’acte d’enseigner influencent leurs pratiques en classe, que les croyances des apprenants influencent leur motivation, leur niveau et leurs efforts déployés pour l’apprentissage de la langue, et qu’une divergence entre les croyances des professeurs et celles des apprenants peut entraîner des effets négatifs sur l’apprentissage de la langue cible, ce qui indique ainsi qu’il est de grande importance d’explorer les croyances. Ainsi, la présente étude vise à explorer les croyances des professeurs égyptiens et celles de leurs étudiants en ce qui a trait à la rétroaction corrective à l’oral, la différence entre ces croyances, et l’identification des pratiques réelles des professeurs afin de décrire à quel point celles-ci reflètent les croyances exprimées. Pour ce faire, un questionnaire a été administré à 175 étudiants et 25 professeurs appartenant à trois universités égyptiennes afin de déterminer leurs croyances déclarées. Des entrevues semi-dirigées et des observations directes ont été réalisées auprès de neuf des 25 professeurs participants pour mieux déterminer leurs croyances et leurs pratiques rétroactives. Les résultats obtenus ont révélé qu’il existe des divergences importantes entre les croyances des professeurs et celles des étudiants, d’un côté, et entre les croyances des professeurs et leur pratique, de l’autre côté. Par exemple, la plupart des étudiants ont déclaré leur opposition à l’utilisation de la reformulation alors que presque la moitié des professeurs ont indiqué être en faveur de cette même technique. Les professeurs ont indiqué que leur choix de techniques rétroactives dépend du type d’erreurs et qu’ils préfèrent inciter les apprenants à s’auto corriger. Cependant, ces mêmes professeurs ont utilisé la reformulation pour corriger la majorité des erreurs de leurs apprenants, quelle que soit leur nature. Nous parvenons ainsi à la conclusion que l’utilisation de la reformulation, qui fait l’objet d’une divergence au niveau des croyances, pourrait être à l’origine du manque de précision langagière rapporté par les chercheurs.

Conception et synthèse d’aminoglycosides semi-synthétiques

Plusieurs aminoglycosides font partie d’une famille d’antibiotiques à large spectre d’action. Les aminoglycosides ayant une activité antibiotique viennent interférer dans la synthèse protéique effectuée par les bactéries. Les protéines mal codées entraineront la mort cellulaire. Au fil des années, de nombreux cas de résistance ont émergé après une utilisation soutenue des aminoglycosides. De nombreux aminoglycosides semi-synthétiques ont été synthétisés avec comme objectif de restaurer leur activité antimicrobienne. Parmi les modifications ayant connu du succès, notons la didésoxygénation d’un diol et l’introduction de la chaine latérale HABA. Des études précédentes ont montré l’efficacité de ces modifications sur les aminoglycosides. Les présents travaux portent sur l’installation de la chaine latérale HABA et la didésoxygénation d’un diol sur la paromomycine et la néomycine. La didésoxygénation sélective des diols a été effectuée en utilisant la méthodologie développée par Garegg et Samuelsson, une variation de la réaction de Tipson-Cohen. Cette méthode a permis l’obtention du motif didésoxygéné sur les cycles A et D dans des rendements jamais égalés pour ce motif synthétique. La chaîne latérale a été introduite en tirant profit de la réactivité et de la sélectivité d’un carbamate cyclique. Ces méthodes combinées ont permis la synthèse efficace de nombreux analogues semi-synthétiques nouveaux. La 3',4'-didéhydro-N-1-HABA-néomycine et la 3',4',3''',4'''-tétradésoxy-N-1-HABA-néomycine montrent une activité impressionnante contre des souches de bactéries résistantes aux aminoglycosides. Des tests de toxicité effectués en collaboration avec Achaogen Inc. ont démontré que ces composés sont relativement toxiques sur les cellules rénales de type H2K, ce qui réduit de façon importante leur index thérapeutique. Afin d’abaisser la toxicité des composés, la relation entre toxicité et basicité a été explorée. Des substitutions de l’amine en 6''' ont été effectuées afin d’abaisser la basicité de l’amine. Les résultats de toxicité et d’activité antimicrobienne démontrent une corrélation importante entre la basicité des amines et la toxicité/activité des aminoglycosides antibiotiques. L’effet d’une modulation du pKa a aussi été exploré en installant des chaines fluorées sur l’amine en 6''' de la paromomycine et de la néomycine. Une séquence synthtétique pour isoler l’amine en 6''' de la néomycine a aussi été développée.

Dynamique de séparation de charges à l'hétérojonction de semi-conducteurs organiques

Une compréhension profonde de la séparation de charge à l’hétérojonction de semi-con- ducteurs organiques est nécessaire pour le développement de diodes photovoltaïques organiques plus efficaces, ce qui serait une grande avancée pour répondre aux besoins mondiaux en énergie durable. L’objectif de cette thèse est de décrire les processus impliqués dans la séparation de charges à hétérojonctions de semi-conducteurs organiques, en prenant en exemple le cas particulier du PCDTBT: PCBM. Nous sondons les excitations d’interface à l’aide de méthodes spectroscopiques résolues en temps couvrant des échelles de temps de 100 femto- secondes à 1 milliseconde. Ces principales méthodes spectroscopiques sont la spectroscopie Raman stimulée femtoseconde, la fluorescence résolue en temps et l’absorption transitoire. Nos résultats montrent clairement que le transfert de charge du PCDTBT au PCBM a lieu avant que l’exciton ne soit relaxé et localisé, un fait expérimental irréconciliable avec la théorie de Marcus semi-classique. La paire de charges qui est créée se divise en deux catégories : les paires de polarons géminales non piégées et les paires profondément piégées. Les premiers se relaxent rapidement vers l’exciton à transfert de charge, qui se recombine radiativement avec une constante de temps de 1– 2 nanoseconde, alors que les seconds se relaxent sur de plus longues échelles de temps via l’effet tunnel. Notre modèle photophysique quantitatif démontre que 2 % de l’excitation créée ne peut jamais se dissocier en porteurs de charge libre, un chiffre qui est en accord avec les rendements élevés rapportés pour ce type de système.