abstract: Models using feature interactions have been applied successfully in many areas such as biomedical analysis, recommender systems. The popularity of using feature interactions mainly lies in (1) they are able to capture the nonlinearity of the data compared with linear effects and (2) they enjoy great interpretability. In this thesis, I propose a series of formulations using feature interactions for real world problems and develop efficient algorithms for solving them. Specifically, I first propose to directly solve the non-convex formulation of the weak hierarchical Lasso which imposes weak hierarchy on individual features and interactions but can only be approximately solved by a convex relaxation in existing studies. I further propose to use the non-convex weak hierarchical Lasso formulation for hypothesis testing on the interaction features with hierarchical assumptions. Secondly, I propose a type of bi-linear models that take advantage of interactions of features for drug discovery problems where specific drug-drug pairs or drug-disease pairs are of interest. These models are learned by maximizing the number of positive data pairs that rank above the average score of unlabeled data pairs. Then I generalize the method to the case of using the top-ranked unlabeled data pairs for representative construction and derive an efficient algorithm for the extended formulation. Last but not least, motivated by a special form of bi-linear models, I propose a framework that enables simultaneously subgrouping data points and building specific models on the subgroups for learning on massive and heterogeneous datasets. Experiments on synthetic and real datasets are conducted to demonstrate the effectiveness or efficiency of the proposed methods. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2018

O trabalho aqui proposto perseguiu a compreensão e a explicação da constituição da produção espacial num dado fragmento da grande metrópole paulistana, a região da Granja Viana pertencente ao município de Cotia. Os espaços foram sendo privatizados e produzidos de forma hierarquizada e os meios que influenciaram e ainda influenciam na constituição e manutenção de lugares nobres e periféricos dentro de uma mesma região estão estritamente ligados a uma estratégia de (re)produção do capital por meio do mercado imobiliário. A Granja Viana \"vendida\" carrega consigo um ideal de qualidade de vida associado ao requinte e segurança que estabelece uma identidade aos empreendimentos oferecidos. Porém existe algo a mais no espaço e que nem sempre é aparente: trata-se da diversidade e dos conflitos. Nesta região o velho e o novo, assim como a abundância e a raridade se mantêm num permanente fazer e refazer de um espaço que não é homogêneo. / This work attempted to make understandable and to explain how a production of space is established in a small portion of a big city. The object of this study is Granja Viana area located in Cotia city, near São Paulo. This area has been privatized and developed in a hierarchical way, and the means that influenced and still influence the building up and the maintenance of noble places and peripheries in a same region are strictly related to a capital grow strategy thru the real estate market. The \"sold\" Granja Viana carries an ideal of quality life associated to refinement and security, which creates an identity to the offered enterprises. But there is something else in this space, which is not always evident that is diversities and conflicts. Old and new in this region as well as abundance and scarcity, maintain an invariable make and remake of a heterogeneous space.

Thesis (M.Sc.(Chemistry)) -- University of Limpopo, 2018. / Refer to the document / NRF-Sasol Inzalo Foundation

H-matrix techniques use a data-sparse tree structure to represent a dense or a sparse matrix. The leaves of the tree store matrix sub-blocks that are represented in full-matrix format or Rk-matrix (low rank matrix) format. H-matrix arithmetic is defined over the H-matrix representation, which includes operations such as addition, multiplication, inversion, and LU factorization. These H-matrix operations approximate results with almost optimal computational complexity. Based on the properties of H-matrices, the H-matrix preconditioner technique has been introduced. It uses H-matrix operations to construct preconditioners, which are used in iterative methods to speed up the solution of large systems of linear equations (Ax = b). To apply the H-matrix preconditioner technique, the first step is to represent a problem in H-matrix format. The approaches to construct an H-matrix can be divided into two categories: geometric approaches and algebraic approaches. In this thesis, we present our contributions to algebraic H-matrix construction approaches and H-matrix preconditioner technique. We have developed a new algebraic H-matrix construction approach based on matrix graphs and multilevel graph clustering approaches. Based on the new construction approach, we have also developed a scheme to build algebraic H-matrix preconditioners for systems of saddle point type. To verify the effectiveness of our new construction approach and H-matrix preconditioner scheme, we have applied them to solve various systems of linear equations arising from finite element methods and meshfree methods. The experimental results show that our preconditioners are competitive to other H-matrix preconditioners based on domain decomposition and existing preconditioners such as JOR and AMG preconditioners. Our H-matrix construction approach and preconditioner technique provide an alternative effective way to solve large systems of linear equations.

Integrating existing models of emotional expressivity, the 3-level hierarchical model contains a general factor of emotional expressivity vs. inexpressivity at the highest level; relatively independent factors of positive and negative expressivity at the second-order level; and discrete expressivity factors of sadness, hostility, guilt/shame, fear, joviality, confidence and amusement at the lowest level. The bottom-up analytic strategy consisted of identifying first the structure of the discrete affects; subsequent second-order factor analyses supported the existence of the higher order factors. The Iowa Scales of Emotional Expressivity (ISEE)--a hierarchical set of scales--systematically incorporate the level of abstraction of the items to assess each level of the hierarchy. Structural analyses replicated across college student (N = 387) and young adult (N = 344) samples with strong comparability coefficients. Striking differences existed in comparisons of the nomological relations of the general factor level vs. second-order level--Positive and Negative Expressivity demonstrated differential relations with Extraversion and Neuroticism and incremental predictive validity beyond Positive and Negative Affect, respectively. The ISEE demonstrated convergent and discriminant validity with existing scales and through multi-trait multi-method analyses of self-other agreement and test-retest data. Although test-retest correlations were less than optimal, the ISEE improve upon existing measures of emotional expressivity by extending the assessment to the discrete affect level and by creating Positive and Negative Expressivity scales with improved discriminant validity and clearer differential relations.

A school district in Utah implemented an instructional coaching program intended to increase student achievement in reading and mathematics. Program administrators wished to determine the degree to which certain elements of instructional coaching (time, activities, context, and content) affected student achievement. Student achievement data were collected using state reading and math assessments; coaching data were collected using coaching time logs; other data were obtained from the district. Data were analyzed to determine which predictors could appropriately be included in a hierarchical linear model (HLM) predicting student achievement. A threelevel fully unconditional model was applied to determine the relative effect of grouped factors at the student, class, and school levels. Approximately 90% of the total variance in student achievement (both explained and unexplained by the model) was observed at the student level. Unconditional growth models were constructed to determine whether student-level factors varied significantly across classes and whether class-level factors varied significantly across schools. Each identified factor was included (as random or fixed) in one of eight explanatory HLMs to measure the effect of specific coaching factors on predicted student achievement. Noncoaching factors were included in the models to reduce extraneous variance and strengthen the models’ ability to describe the effect of coaching factors. Inclusion of factors reduced unexplained student-level variance by approximately 45% in the language arts models and 54% in the math models. There was no evidence that coaching time had a direct effect on student achievement. Some of the coaching activities, contexts, and contents did affect predicted achievement significantly. This report outlines those observed effects in detail. The most notable finding was that students in classrooms where coaches spend more time conferencing with teachers about student achievement data had higher predicted scores. Due to the nature of the dependent variable (achievement) and inherent methodological challenges associated with measuring the effect of class-level interventions, effect sizes observed in this study were relatively small. The resulting recommendations for practice were that coaches focus less on the quantity of time they spend with teachers and more on selecting activities, context, and content that are likely to yield the greatest results.

The thesis reviews current admission control schemes in cellular IP networks. It proposes an improved version of Threshold Access Sharing and a new scheme: weight-based scheme. Finally, an admission control scheme for hierarchical cellular network is introduced.

<p>This thesis looks into how one could use Hierarchal Temporal Memory (HTM) networks to generate models that could be used as trading algorithms. The thesis begins with a brief introduction to algorithmic trading and commonly used concepts when developing trading algorithms. The thesis then proceeds to explain what an HTM is and how it works. To explore whether an HTM could be used to generate models that could be used as trading algorithms, the thesis conducts a series of experiments. The goal of the experiments is to iteratively optimize the settings for an HTM and try to generate a model that when used as a trading algorithm would have more profitable trades than losing trades. The setup of the experiments is to train an HTM to predict if it is a good time to buy some shares in a security and hold them for a fixed time before selling them again. A fair amount of the models generated during the experiments was profitable on data the model have never seen before, therefore the author concludes that it is possible to train an HTM so it can be used as a profitable trading algorithm.</p>

Most existing Peer-to-Peer (P2P) systems support only title-based searches and are limited in functionality when compared to today’s search engines. In this paper, we present the design of a distributed P2P information sharing system that supports semantic-based content searches of relevant documents. First, we propose a general and extensible framework for searching similar documents in P2P network. The framework is based on the novel concept of Hierarchical Summary Structure. Second, based on the framework, we develop our efficient document searching system, by effectively summarizing and maintaining all documents within the network with different granularity. Finally, an experimental study is conducted on a real P2P prototype, and a large-scale network is further simulated. The results show the effectiveness, efficiency and scalability of the proposed system. / Singapore-MIT Alliance (SMA)

The visual recognition of complex movements and actions is crucial for communication and survival in many species. Remarkable sensitivity and robustness of biological motion perception have been demonstrated in psychophysical experiments. In recent years, neurons and cortical areas involved in action recognition have been identified in neurophysiological and imaging studies. However, the detailed neural mechanisms that underlie the recognition of such complex movement patterns remain largely unknown. This paper reviews the experimental results and summarizes them in terms of a biologically plausible neural model. The model is based on the key assumption that action recognition is based on learned prototypical patterns and exploits information from the ventral and the dorsal pathway. The model makes specific predictions that motivate new experiments.