No description available.

In drug discovery, thousands of compounds are assayed to detect activity against a biological target. The goal of drug discovery is to identify compounds that are active against the target (e.g. inhibit a virus). Statistical learning in drug discovery seeks to build a model that uses descriptors characterizing molecular structure to predict biological activity. However, the characteristics of drug discovery data can make it difficult to model the relationship between molecular descriptors and biological activity. Among these characteristics are the rarity of active compounds, the large volume of compounds tested by high-throughput screening, and the complexity of molecular structure and its relationship to activity. This thesis focuses on the design of statistical learning algorithms/models and their applications to drug discovery. The two main parts of the thesis are: an algorithm-based statistical method and a more formal model-based approach. Both approaches can facilitate and accelerate the process of developing new drugs. A unifying theme is the use of unsupervised methods as components of supervised learning algorithms/models. In the first part of the thesis, we explore a sequential screening approach, Cluster Structure-Activity Relationship Analysis (CSARA). Sequential screening integrates High Throughput Screening with mathematical modeling to sequentially select the best compounds. CSARA is a cluster-based and algorithm driven method. To gain further insight into this method, we use three carefully designed experiments to compare predictive accuracy with Recursive Partitioning, a popular structureactivity relationship analysis method. The experiments show that CSARA outperforms Recursive Partitioning. Comparisons include problems with many descriptor sets and situations in which many descriptors are not important for activity. In the second part of the thesis, we propose and develop constrained mixture discriminant analysis (CMDA), a model-based method. The main idea of CMDA is to model the distribution of the observations given the class label (e.g. active or inactive class) as a constrained mixture distribution, and then use Bayes’ rule to predict the probability of being active for each observation in the testing set. Constraints are used to deal with the otherwise explosive growth of the number of parameters with increasing dimensionality. CMDA is designed to solve several challenges in modeling drug data sets, such as multiple mechanisms, the rare target problem (i.e. imbalanced classes), and the identification of relevant subspaces of descriptors (i.e. variable selection). We focus on the CMDA1 model, in which univariate densities form the building blocks of the mixture components. Due to the unboundedness of the CMDA1 log likelihood function, it is easy for the EM algorithm to converge to degenerate solutions. A special Multi-Step EM algorithm is therefore developed and explored via several experimental comparisons. Using the multi-step EM algorithm, the CMDA1 model is compared to model-based clustering discriminant analysis (MclustDA). The CMDA1 model is either superior to or competitive with the MclustDA model, depending on which model generates the data. The CMDA1 model has better performance than the MclustDA model when the data are high-dimensional and unbalanced, an essential feature of the drug discovery problem! An alternate approach to the problem of degeneracy is penalized estimation. By introducing a group of simple penalty functions, we consider penalized maximum likelihood estimation of the CMDA1 and CMDA2 models. This strategy improves the convergence of the conventional EM algorithm, and helps avoid degenerate solutions. Extending techniques from Chen et al. (2007), we prove that the PMLE’s of the two-dimensional CMDA1 model can be asymptotically consistent.

In drug discovery, thousands of compounds are assayed to detect activity against a biological target. The goal of drug discovery is to identify compounds that are active against the target (e.g. inhibit a virus). Statistical learning in drug discovery seeks to build a model that uses descriptors characterizing molecular structure to predict biological activity. However, the characteristics of drug discovery data can make it difficult to model the relationship between molecular descriptors and biological activity. Among these characteristics are the rarity of active compounds, the large volume of compounds tested by high-throughput screening, and the complexity of molecular structure and its relationship to activity. This thesis focuses on the design of statistical learning algorithms/models and their applications to drug discovery. The two main parts of the thesis are: an algorithm-based statistical method and a more formal model-based approach. Both approaches can facilitate and accelerate the process of developing new drugs. A unifying theme is the use of unsupervised methods as components of supervised learning algorithms/models. In the first part of the thesis, we explore a sequential screening approach, Cluster Structure-Activity Relationship Analysis (CSARA). Sequential screening integrates High Throughput Screening with mathematical modeling to sequentially select the best compounds. CSARA is a cluster-based and algorithm driven method. To gain further insight into this method, we use three carefully designed experiments to compare predictive accuracy with Recursive Partitioning, a popular structureactivity relationship analysis method. The experiments show that CSARA outperforms Recursive Partitioning. Comparisons include problems with many descriptor sets and situations in which many descriptors are not important for activity. In the second part of the thesis, we propose and develop constrained mixture discriminant analysis (CMDA), a model-based method. The main idea of CMDA is to model the distribution of the observations given the class label (e.g. active or inactive class) as a constrained mixture distribution, and then use Bayes’ rule to predict the probability of being active for each observation in the testing set. Constraints are used to deal with the otherwise explosive growth of the number of parameters with increasing dimensionality. CMDA is designed to solve several challenges in modeling drug data sets, such as multiple mechanisms, the rare target problem (i.e. imbalanced classes), and the identification of relevant subspaces of descriptors (i.e. variable selection). We focus on the CMDA1 model, in which univariate densities form the building blocks of the mixture components. Due to the unboundedness of the CMDA1 log likelihood function, it is easy for the EM algorithm to converge to degenerate solutions. A special Multi-Step EM algorithm is therefore developed and explored via several experimental comparisons. Using the multi-step EM algorithm, the CMDA1 model is compared to model-based clustering discriminant analysis (MclustDA). The CMDA1 model is either superior to or competitive with the MclustDA model, depending on which model generates the data. The CMDA1 model has better performance than the MclustDA model when the data are high-dimensional and unbalanced, an essential feature of the drug discovery problem! An alternate approach to the problem of degeneracy is penalized estimation. By introducing a group of simple penalty functions, we consider penalized maximum likelihood estimation of the CMDA1 and CMDA2 models. This strategy improves the convergence of the conventional EM algorithm, and helps avoid degenerate solutions. Extending techniques from Chen et al. (2007), we prove that the PMLE’s of the two-dimensional CMDA1 model can be asymptotically consistent.

Bibliotecas digitais tornaram-se uma importante fonte de informação para comunidades científicas. Entretanto, por coletar dados de diferentes fontes, surge o problema de informações ambíguas ou duplicadas de nomes de autores. Métodos tradicionais de desambiguação de nomes utilizam informação sintática de atributos. Todavia, recentemente o uso de redes de relacionamentos, que traz informação semântica, tem sido estudado em desambiguação de dados. Em desambiguação de nomes de autores, relações de co-autoria podem ser usadas para criar uma rede social, que pode ser utilizada para melhorar métodos de desambiguação de nomes de autores. Esta dissertação apresenta um estudo do impacto de adicionar análise de redes sociais a métodos de desambiguação de nomes de autores baseados em informação sintática de atributos. Nós apresentamos uma abordagem de aprendizagem de máquina baseada em Programação Genética e a utilizamos para avaliar o impacto de adicionar análise de redes sociais a desambiguação de nomes de autores. Através de experimentos usando subconjuntos de bibliotecas digitais reais, nós demonstramos que o uso de análise de redes sociais melhora de forma significativa a qualidade dos resultados. Adicionalmente, nós demonstramos que as funções de casamento criadas por nossa abordagem baseada em Programação Genética são capazes de competir com métodos do estado da arte. / Digital libraries have become an important source of information for scientific communities. However, by gathering data from different sources, the problem of duplicate and ambiguous information about author names arises. Traditional methods of name disambiguation use syntactic attribute information. However, recently the use of relationship networks, which provides semantic information, has been studied in data disambiguation. In author name disambiguation, the co-authorship relations can be used to create a social network, which can be used to improve author name disambiguation methods. This dissertation presents a study of the impact of adding social network analysis to author name disambiguation methods based on syntactic attribute information. We present a machine learning approach based on Genetic Programming and use it to evaluate the impact of social network analysis in author name disambiguation. Through experiments using subsets of real digital libraries, we show that the use of social network analysis significantly improves the quality of results. Also, we demonstrate that match functions created by our Genetic Programming approach are able to compete with state-of-the-art methods.

Bibliotecas digitais tornaram-se uma importante fonte de informação para comunidades científicas. Entretanto, por coletar dados de diferentes fontes, surge o problema de informações ambíguas ou duplicadas de nomes de autores. Métodos tradicionais de desambiguação de nomes utilizam informação sintática de atributos. Todavia, recentemente o uso de redes de relacionamentos, que traz informação semântica, tem sido estudado em desambiguação de dados. Em desambiguação de nomes de autores, relações de co-autoria podem ser usadas para criar uma rede social, que pode ser utilizada para melhorar métodos de desambiguação de nomes de autores. Esta dissertação apresenta um estudo do impacto de adicionar análise de redes sociais a métodos de desambiguação de nomes de autores baseados em informação sintática de atributos. Nós apresentamos uma abordagem de aprendizagem de máquina baseada em Programação Genética e a utilizamos para avaliar o impacto de adicionar análise de redes sociais a desambiguação de nomes de autores. Através de experimentos usando subconjuntos de bibliotecas digitais reais, nós demonstramos que o uso de análise de redes sociais melhora de forma significativa a qualidade dos resultados. Adicionalmente, nós demonstramos que as funções de casamento criadas por nossa abordagem baseada em Programação Genética são capazes de competir com métodos do estado da arte. / Digital libraries have become an important source of information for scientific communities. However, by gathering data from different sources, the problem of duplicate and ambiguous information about author names arises. Traditional methods of name disambiguation use syntactic attribute information. However, recently the use of relationship networks, which provides semantic information, has been studied in data disambiguation. In author name disambiguation, the co-authorship relations can be used to create a social network, which can be used to improve author name disambiguation methods. This dissertation presents a study of the impact of adding social network analysis to author name disambiguation methods based on syntactic attribute information. We present a machine learning approach based on Genetic Programming and use it to evaluate the impact of social network analysis in author name disambiguation. Through experiments using subsets of real digital libraries, we show that the use of social network analysis significantly improves the quality of results. Also, we demonstrate that match functions created by our Genetic Programming approach are able to compete with state-of-the-art methods.

Bibliotecas digitais tornaram-se uma importante fonte de informação para comunidades científicas. Entretanto, por coletar dados de diferentes fontes, surge o problema de informações ambíguas ou duplicadas de nomes de autores. Métodos tradicionais de desambiguação de nomes utilizam informação sintática de atributos. Todavia, recentemente o uso de redes de relacionamentos, que traz informação semântica, tem sido estudado em desambiguação de dados. Em desambiguação de nomes de autores, relações de co-autoria podem ser usadas para criar uma rede social, que pode ser utilizada para melhorar métodos de desambiguação de nomes de autores. Esta dissertação apresenta um estudo do impacto de adicionar análise de redes sociais a métodos de desambiguação de nomes de autores baseados em informação sintática de atributos. Nós apresentamos uma abordagem de aprendizagem de máquina baseada em Programação Genética e a utilizamos para avaliar o impacto de adicionar análise de redes sociais a desambiguação de nomes de autores. Através de experimentos usando subconjuntos de bibliotecas digitais reais, nós demonstramos que o uso de análise de redes sociais melhora de forma significativa a qualidade dos resultados. Adicionalmente, nós demonstramos que as funções de casamento criadas por nossa abordagem baseada em Programação Genética são capazes de competir com métodos do estado da arte. / Digital libraries have become an important source of information for scientific communities. However, by gathering data from different sources, the problem of duplicate and ambiguous information about author names arises. Traditional methods of name disambiguation use syntactic attribute information. However, recently the use of relationship networks, which provides semantic information, has been studied in data disambiguation. In author name disambiguation, the co-authorship relations can be used to create a social network, which can be used to improve author name disambiguation methods. This dissertation presents a study of the impact of adding social network analysis to author name disambiguation methods based on syntactic attribute information. We present a machine learning approach based on Genetic Programming and use it to evaluate the impact of social network analysis in author name disambiguation. Through experiments using subsets of real digital libraries, we show that the use of social network analysis significantly improves the quality of results. Also, we demonstrate that match functions created by our Genetic Programming approach are able to compete with state-of-the-art methods.

Yes / Many in-vitro methods for prediction of skin permeability have been reported in literature. Cerasome electrokinetic chromatography is one of the most sophisticated approaches representing a maximum level of similarity to the lipid phase of the stratum corneum. One goal of this study was to investigate the affinity pattern of Cerasome and to compare it with the permeability profile of human skin. Another purpose was to study the applicability of Hansen solubility parameters for modelling skin permeation and to investigate the predictive and explanatory potential of this method. Visualisation in Hansen diagrams revealed very similar profiles of Cerasome electrokinetic chromatography retention factors and skin permeability coefficients. In both cases, the characteristic pattern with two clusters of highly retained or highly permeable substances could be shown to be mainly caused by two groups of compounds, one of them with high affinity to ceramides, fatty acids and lecithin and the other being more affine to cholesterol. If based on a sufficiently comprehensive experimental dataset, model-independent predictions of skin permeability data using three-component Hansen solubility parameters are able to achieve similar accuracy as calculations made with an Abraham linear free energy relationship model in which the compounds are characterized by seven physicochemical descriptors.

No description available.

Магистерская диссертация выполнена на 135 листах. Количество таблиц 24 (без приложений), рисунков 11 (без приложений), формул 10 (без приложений), источников 79, приложений 3. Цель исследования состоит в развитии методических основ исследования корпоративной социальной ответственности в качестве инструмента повышения уровня конкурентоспособности предприятий малого и среднего бизнеса. Задачи исследования: - изучить теоретические и методические вопросы формирования конкурентоспособности предприятий малого и среднего бизнеса; - исследовать специфические особенности реализации корпоративной социальной ответственности на предприятиях малого и среднего бизнеса и охарактеризовать такие практики в качестве инструмента повышения уровня конкурентоспособности; - разработать методику оценки влияния практик корпоративной социальной ответственности предприятий малого и среднего бизнеса на уровень их конкурентоспособности; - апробировать разработанную методику и определить направления ее возможного совершенствования. Новизна исследования состоит в следующих аспектах: - разработан концептуальный подход к исследованию корпоративной социальной ответственности как инструмента повышения конкурентоспособности предприятий малого и среднего бизнеса, включающий в себя: выделение наиболее значимых субъектов конкурентоспособности, на которых могут быть направлены практики корпоративной социальной ответственности; формирование набора экономико-статистических индикаторов, позволяющих оценивать влияние практик корпоративной социальной ответственности на конкурентоспособность предприятия в разрезе каждого выделенного субъекта; описание механизма (способа) воздействия практик корпоративной социальной ответственности на каждого выделенного субъекта конкурентоспособности; определение перечня возможных положительных эффектов, возникающих вследствие воздействия практик корпоративной социальной ответственности на каждого субъекта конкурентоспособности; формализацию процесса оценки положительных эффектов на базе статистических методов, что развивает инструментально-методические основы исследования конкурентоспособности и положения теории управления предприятием; - предложена методика исследования влияния практик корпоративной социальной ответственности, реализуемых в отношении сотрудников предприятий малого и среднего бизнеса, на уровень их конкурентоспособности, предполагающая выделение и систематизацию показателей, способов и эффектов от влияния таких практик на уровень конкурентоспособности; разработку инструментария исследования на основе анкетного опроса сотрудников; определение этапов и организационных параметров исследования, что позволяет получать количественные (статистические) оценки влияния таких практик на уровень конкурентоспособности и формировать информационно-аналитический базис обоснования необходимости более широкого внедрения корпоративной социальной ответственности на предприятиях малого и среднего бизнеса. Практическая значимость состоит в возможности применения разработанной методики предприятиями малого и среднего бизнеса для оценки степени влияния практик корпоративной социальной ответственности на уровень конкурентоспособности бизнеса, что создает информационно-аналитическую основу для принятия более эффективных управленческих решений и выработки эффективных стратегий и тактик дальнейшего развития предприятия. / The master's thesis is made on 132 sheets. The number of tables 23 (without appendices), figures 11 (without appendices), formulas 10 (without appendices), sources 79, appendices 3. The purpose of the study is to develop the methodological foundations of the study of corporate social responsibility as a tool for improving the competitiveness of small and medium-sized businesses. Objectives of the study: - to study the theoretical and methodological issues of forming the competitiveness of small and medium-sized businesses; - to study the specific features of the implementation of corporate social responsibility in small and medium-sized businesses and to characterize such practices as a tool for improving the level of competitiveness; - develop a methodology for assessing the impact of corporate social responsibility practices of small and medium-sized businesses on their competitiveness; - to test the developed methodology and determine the directions of its possible improvement. The novelty of the research consists in the following aspects: - developed a conceptual approach to the study of corporate social responsibility as a tool to enhance the competitiveness of small and medium-sized businesses, including: selection of the most important subjects of competitiveness, which can be directed to the practices of corporate social responsibility; the formation of a set of economic and statistical indicators to assess the impact of CSR practices on company competitiveness in the context of each selected entity; the description of the mechanism (ways) the impact of CSR practices on each selected entity competitiveness; the determination of the list of possible positive effects arising from the impact of CSR practices on every subject of competitiveness; the formalisation of the process of assessing the positive effects on the basis of statistical methods that develops instrumental and methodological basis of the study of competitiveness and the theory of enterprise management; - a methodology for studying the impact of corporate social responsibility practices implemented in relation to employees of small and medium-sized businesses on the level of their competitiveness is proposed, which involves the identification and systematization of indicators, methods and effects of the impact of such practices on the level of competitiveness; the development of research tools based on a questionnaire survey of employees; determination of the stages and organizational parameters of the study, which allows us to obtain quantitative (statistical) assessments of the impact of such practices on the level of competitiveness and to form an information and analytical basis for justifying the need for a broader introduction of corporate social responsibility in small and medium-sized businesses. The practical significance lies in the possibility of using the developed methodology by small and medium-sized businesses to assess the degree of influence of corporate social responsibility practices on the level of business competitiveness, which creates an information and analytical basis for making more effective management decisions and developing effective strategies and tactics for further development of the enterprise.