In this thesis we implement an ensemble of sequence analysis strategies aimed at identifying functional and structural protein features. The first part of this work was dedicated to two case studies of specific proteins analyzed to provide candidate functional positions for experimental validation: the protein alpha-synuclein (αsyn) and the alanine racemases protein family. In the case of αsyn, the objective was to predict its aggregation prone regions. For the alanine racemase protein family, the scope was to predict sites responsible for substrate specificity. In these two studies, computational predictions allowed systematically exploring potentially functionally relevant protein sites in an efficient manner that may not be possible to implement with traditional experimental approaches. Our strategy provided a powerful forecasting tool for the selection of candidate sites to be later verified experimentally. In the second part, we analyze the role of intrinsic disorder (ID) as a modulator of protein function in different organisms and cellular processes, which is largely unexplored. As key components of the diverse cellular pathways, disordered proteins are often involved in many diseases, including cancer and neurodegenerative diseases. Thus, there is an impeding need to unveil the general principles underlying the role of ID in proteins. We provide a multi-scale analysis of the involvement of ID in protein function starting with a large-scale analysis at genomic level of the role of ID in Arabidopsis, zooming in into the specific processes of vesicular trafficking in Human and yeast, and finally focusing on specific proteins of diverse organisms. The results of this thesis provide a better understanding of the functional roles mediated by ID in different organisms and biological processes, such as acting as flexible linkers connecting structured domains, mediating protein-protein interactions, and assisting the quick assembly of large macromolecular complexes. In addition, we present evidence of the use of ID as a mechanism to increase the complexity of protein and biological networks, and as a means to increase the adaptability of proteins in specific processes. Thus, our results contribute to elucidating the relationship between network and organismal complexity and ID, while they also provide evidence of the evolutionary advantages offered by ID.

The outsourcing of the business logistics activities have became a trend. The goals of my research want to understand that: (1) The outsourcing extent of the logistics activities. (2) The logistics outsourcing considered factors. (3) How the firm size affect the (1) and (2). (4) How the logistics network complexity affect the (1) and (2). The research¡¯s samples are the PC and peripheral equipment companies in Taiwan. The capital, the revenue, and the number of the employee measure the size of a firm. The logistics network complexity is measured by the number of the customers, the number of the suppliers, the number of the countries that customers in, and the number of the countries that the suppliers in. The result shows that the companies think the cost saving factors is most important when they decide to outsource the activity. The great extent of the traffic transportation, salvage and scrap disposal and return goods handling are outsourced. The companies whose sizes are small put great emphasis on the ¡°logistics information systems¡± factors. The big companies have a tendency to outsource the physical moved activities like traffic transportation, salvage and scrap disposal. The small companies are apt to outsource the process handling activities. In the logistics network complexity, the suppliers make greater difference in the outsourcing logistics activities than the customers. The numbers of the countries affect the logistics outsourcing considered factors.

Machine learning (ML) is an area of computer science that gives computers the ability to learn data patterns without prior programming for those patterns. Using neural networks in this area is based on simulating the biological functions of neurons in brains to learn patterns in data, giving computers a predictive ability to comprehend how data can be clustered. This research investigates the possibilities of using neural networks for classifying email, i.e. working as an email case manager. A Deep Neural Network (DNN) are multiple layers of neurons connected to each other by trainable weights. The main objective of this thesis was to evaluate how the three input arguments - data size, training time and neural network structure – affects the accuracy of Deep Neural Networks pattern recognition; also an evaluation of how the DNN performs compared to the statistical ML method, Naïve Bayes, in the form of prediction accuracy and complexity; and finally the viability of the resulting DNN as a case manager. Results show an improvement of accuracy on our networks with the increase of training time and data size respectively. By testing increasingly complex network structures (larger networks of neurons with more layers) it is observed that overfitting becomes a problem with increased training time, i.e. how accuracy decrease after a certain threshold of training time. Naïve Bayes classifiers performs worse than DNN in terms of accuracy, but better in reduced complexity; making NB viable on mobile platforms. We conclude that our developed prototype may work well in tangent with existing case management systems, tested by future research.

Learning with neural networks depends on the particular parametrization of the functions represented by the network, that is, the assignment of parameters to functions. It also depends on the identity of the functions, which get assigned typical parameters at initialization, and, later, the parameters that arise during training. The choice of the activation function is a critical aspect of the network design that influences these function properties and requires investigation. This thesis focuses on analyzing the expected behavior of networks with maxout (multi-argument) activation functions. On top of enhancing the practical applicability of maxout networks, these findings add to the theoretical exploration of activation functions beyond the common choices. We believe this work can advance the study of activation functions and complicated neural network architectures. We begin by taking the number of activation regions as a complexity measure and showing that the practical complexity of deep networks with maxout activation functions is often far from the theoretical maximum. This analysis extends the previous results that were valid for deep neural networks with single-argument activation functions such as ReLU. Additionally, we demonstrate that a similar phenomenon occurs when considering the decision boundaries in classification tasks. We also show that the parameter space has a multitude of full-dimensional regions with widely different complexity and obtain nontrivial lower bounds on the expected complexity. Finally, we investigate different parameter initialization procedures and show that they can increase the speed of the gradient descent convergence in training. Further, continuing the investigation of the expected behavior, we study the gradients of a maxout network with respect to inputs and parameters and obtain bounds for the moments depending on the architecture and the parameter distribution. We observe that the distribution of the input-output Jacobian depends on the input, which complicates a stable parameter initialization. Based on the moments of the gradients, we formulate parameter initialization strategies that avoid vanishing and exploding gradients in wide networks. Experiments with deep fully-connected and convolutional networks show that this strategy improves SGD and Adam training of deep maxout networks. In addition, we obtain refined bounds on the expected number of linear regions, results on the expected curve length distortion, and results on the NTK. As the result of the research in this thesis, we develop multiple experiments and helpful components and make the code for them publicly available.

As startups digitais ocorrem como um fenômeno da própria Sociedade em Rede e de uma mudança profunda e qualitativa nas relações entre pessoas, espaços e coisas. Seus fluxos produtivos traduzem uma intensa transitividade dos espaços de produção, um protagonismo permanente com novas tecnologias digitais e uma mobilização do corpo como máquina produtiva. Esta pesquisa aborda a criação das startups na sua fase inicial, ou seja, nos momentos de idealizações do pensamento que age, na relação entre ideia e ação, especificando o movimento de travessia dos seus criadores, aqui chamados de Novos Idealistas, num percurso de alta criticidade denominado Zona Cinzenta. A partir dos pontos de ideação-previsão e experimentação-criação, a pesquisa produz reflexos sobre a arte desse empreendedorismo em situação-embrião. Com o objetivo de produzir um tecimento das suas ações na travessia, a pesquisa realiza um acompanhamento de processos nos territórios moventes das startups, desenvolvendo uma cartografia de conceitos que pulsam com intensidade neste transcurso. As teorias que situam o contexto desse novo modo de produzir, configurado pela hegemonização do trabalho imaterial, referenciam os caminhos da pesquisa, situando o conhecimento como principal força produtiva. Para entender esta travessia, a pesquisa navegou nos fazeres das startups, observando de dentro os movimentos de produção, a linguagem e o acionamento das máquinas comunicativas. Ao acompanhar os desejos e expor cases das ações, esta investigação assinala o fenômeno das startups como um novo empreendedorismo marcadamente digital, ecossistêmico e comunicativo. / Digital startups occur as phenomena of the Internet Society and a deep and qualitative change in relationships between people, places and things. Their production flows reveal an intense transitivity of production spaces, a permanent leading role with new digital technologies, and mobilization of the body as a productive machine. This research deals with the creation of startups in their initial phase. That is, in those moments of idealization of thoughts that relate idea and action, that specify the movement of their creators, here called New Idealists, on a high criticality path through what is called a Grey zone. Starting from the points of ideation-prediction and experimentation-creation, the research produces reflections on the art of this entrepreneurship in an embryo situation. With the goal of interweaving their actions in this passage, the research performs a monitoring process in areas where startups act, mapping concepts that pulsate with intensity along this course. The theories that contextualize this new production mode, in which immaterial labor dominates, reference the paths of research, placing knowledge as the main productive force. To understand this passage, the research navigated the doings of startups, observing from inside the movements of production, the language, the activation of the communication machines. In following the desires and exposing \'cases\' of actions, this research points to the phenomenon of startups as a new entrepreneurship, markedly digital, ecosystemic and communicative.

As startups digitais ocorrem como um fenômeno da própria Sociedade em Rede e de uma mudança profunda e qualitativa nas relações entre pessoas, espaços e coisas. Seus fluxos produtivos traduzem uma intensa transitividade dos espaços de produção, um protagonismo permanente com novas tecnologias digitais e uma mobilização do corpo como máquina produtiva. Esta pesquisa aborda a criação das startups na sua fase inicial, ou seja, nos momentos de idealizações do pensamento que age, na relação entre ideia e ação, especificando o movimento de travessia dos seus criadores, aqui chamados de Novos Idealistas, num percurso de alta criticidade denominado Zona Cinzenta. A partir dos pontos de ideação-previsão e experimentação-criação, a pesquisa produz reflexos sobre a arte desse empreendedorismo em situação-embrião. Com o objetivo de produzir um tecimento das suas ações na travessia, a pesquisa realiza um acompanhamento de processos nos territórios moventes das startups, desenvolvendo uma cartografia de conceitos que pulsam com intensidade neste transcurso. As teorias que situam o contexto desse novo modo de produzir, configurado pela hegemonização do trabalho imaterial, referenciam os caminhos da pesquisa, situando o conhecimento como principal força produtiva. Para entender esta travessia, a pesquisa navegou nos fazeres das startups, observando de dentro os movimentos de produção, a linguagem e o acionamento das máquinas comunicativas. Ao acompanhar os desejos e expor cases das ações, esta investigação assinala o fenômeno das startups como um novo empreendedorismo marcadamente digital, ecossistêmico e comunicativo. / Digital startups occur as phenomena of the Internet Society and a deep and qualitative change in relationships between people, places and things. Their production flows reveal an intense transitivity of production spaces, a permanent leading role with new digital technologies, and mobilization of the body as a productive machine. This research deals with the creation of startups in their initial phase. That is, in those moments of idealization of thoughts that relate idea and action, that specify the movement of their creators, here called New Idealists, on a high criticality path through what is called a Grey zone. Starting from the points of ideation-prediction and experimentation-creation, the research produces reflections on the art of this entrepreneurship in an embryo situation. With the goal of interweaving their actions in this passage, the research performs a monitoring process in areas where startups act, mapping concepts that pulsate with intensity along this course. The theories that contextualize this new production mode, in which immaterial labor dominates, reference the paths of research, placing knowledge as the main productive force. To understand this passage, the research navigated the doings of startups, observing from inside the movements of production, the language, the activation of the communication machines. In following the desires and exposing \'cases\' of actions, this research points to the phenomenon of startups as a new entrepreneurship, markedly digital, ecosystemic and communicative.

Complexity science has come into the limelight in recent years as the scientific community begins to grapple with higher-order natural phenomena that cannot be fully explained via the behaviour of components at lower levels of organization. Network modeling and analysis, being a powerful tool that can capture the interconnections that embody complex behaviour, has therefore been at the forefront of complexity science. In ecology, the network paradigm is relatively young and there remain limitations in many ecological network studies, such as modeling only one type of species interaction at a time, lack of realistic network structure, or non-inclusion of community dynamics and environmental stochasticity. I introduce bioenergetic network models that bring together for the first time many of the fundamental structures and mechanisms of species interactions present in real ecological communities. I then use these models to address some outstanding questions that are relevant to understanding ecological networks at the systems level rather than at the level of subsets of interactions. Firstly, I find that realistic red-shifted environmental noise, and synchrony of species responses to noise, are associated with increased variability in ecosystem properties, with implications for predictive ecological modeling which usually assumes white noise. Next, I look at simultaneous species extinction and invasion, finding that as their individual impacts increase, their combined impact becomes decreasingly additive. In addition, the greater the impact of extinction or invasion, the lesser their reversibility via reintroduction or eradication of the species in question. For modifications of pairwise species interactions by third-party species, a phenomenon that has so far been studied one interaction at a time, I find that the many interaction modifications that occur concurrently in a community can collectively have systematic effects on total biomass and species evenness. Finally, examining a higher level of organization in the form of compartmentalized networks, I find that the relationship between intercompartment connectivity and the impacts of species decline depends considerably on network topology and whether the consumer-resource functional response is prey- or ratio-dependent. Overall, the results vary considerably across model communities with different parameterizations, underscoring the contingency and context dependence of nature that scientists and policy makers alike should no longer ignore. This work hopes to contribute to a growing multidisciplinary understanding, appreciation and management of complex systems that is fundamentally transforming the modern world and giving us insights on how to live more harmoniously within our environment.

No description available.