<div>It is a central problem in both statistics and computer science to understand the theoretical foundation of machine learning, especially deep learning. During the past decade, deep learning has achieved remarkable successes in solving many complex artificial intelligence tasks. The aim of this dissertation is to understand deep neural networks (DNNs) and other nonparametric methods in machine learning. In particular, three machine learning models have been studied: weight normalized DNNs, sparse DNNs, and the compositional nonparametric model.</div><div></div><div><br></div><div>The first chapter presents a general framework for norm-based capacity control for <i>L_p,q</i> weight normalized DNNs. We establish the upper bound on the Rademacher complexities of this family. Especially, with an <i>L_1,infty</i> normalization, we discuss properties of a width-independent capacity control, which only depends on the depth by a square root term. Furthermore, if the activation functions are anti-symmetric, the bound on the Rademacher complexity is independent of both the width and the depth up to a log factor. In addition, we study the weight normalized deep neural networks with rectified linear units (ReLU) in terms of functional characterization and approximation properties. In particular, for an <i>L_1,infty</i> weight normalized network with ReLU, the approximation error can be controlled by the <i>L₁</i> norm of the output layer.</div><div></div><div><br></div><div>In the second chapter, we study <i>L_1,infty</i>-weight normalization for deep neural networks with bias neurons to achieve the sparse architecture. We theoretically establish the generalization error bounds for both regression and classification under the <i>L_1,infty</i>-weight normalization. It is shown that the upper bounds are independent of the network width and <i>k^1/2</i>-dependence on the network depth <i>k</i>. These results provide theoretical justifications on the usage of such weight normalization to reduce the generalization error. We also develop an easily implemented gradient projection descent algorithm to practically obtain a sparse neural network. We perform various experiments to validate our theory and demonstrate the effectiveness of the resulting approach.</div><div></div><div><br></div><div>In the third chapter, we propose a compositional nonparametric method in which a model is expressed as a labeled binary tree of <i>2k+1</i> nodes, where each node is either a summation, a multiplication, or the application of one of the <i>q</i> basis functions to one of the <i>m₁</i> covariates. We show that in order to recover a labeled binary tree from a given dataset, the sufficient number of samples is <i>O(k </i>log<i>(m₁q)+</i>log<i>(k!))</i>, and the necessary number of samples is <i>Omega(k </i>log<i>(m₁q)-</i>log<i>(k!))</i>. We further propose a greedy algorithm for regression in order to validate our theoretical findings through synthetic experiments.</div>

[pt] É crescente na Educação básica a preocupação em formar olhares capazes de aceitar a complexidade das situações reais, sem reduzi-la, sintetizá-la ou livrá-la de suas contradições. Neste contexto, a ideia de preparar alunos competentes para resolverem problemas é recorrente em documentos acerca dos novos objetivos da Educação contemporânea, especialmente nos parâmetros do PISA, o importante programa de avaliação internacional de estudantes, no qual assume o status de símbolo da capacidade de mobilizar conhecimentos para lidar com a realidade. A partir da consideração de que o conceito de resolução de problemas ainda é interpretado de formas bastante díspares, esta dissertação visa explorar possíveis contribuições do campo do Design para a Educação, no sentido de trazer novas perspectivas sobre o seu papel na capacidade dos alunos em lidar com a complexidade do mundo, tal como descrita por Morin. Para tanto, foi traçado um percurso de discussões epistemológicas e metodológicas, desde Simon até Coyne e Findeli, passando por Rittel E Webber, Schon e Buchanan, nas quais é questionada a congruência dos termos problema e solução com a prática e o ensino do campo. Também foram conduzidas entrevistas com professores da Educação básica que cursaram a graduação em Design, como forma de nos engajarmos em um diálogo, do qual poderiam emergir conceitos e conexões que revelassem como os entendimentos próprios do campo do Design poderiam oferecer contribuições para o âmbito educacional. Entendemos que o contexto de propensões e motivações do Design, pela pluralidade de sua dimensão cultural, favorece a transcendência das dicotomias e a compreensão do aspecto fundamental da complexidade. / [en] There is a growing concern in general Education with developing an acceptance of the complexity of real-life situations, without reducing it, synthesizing it or exempting it from its contradictions. In this context, the idea of preparing students to be competent problem solvers is recurring in documents about the new of goals of contemporary Education, especially in the PISA (Programme for International Student Assessment) framework, where it symbolizes the capacity of using multiple understandings to deal with reality. Considering that the concept of problem solving is still interpreted in very unequal ways, this dissertation intends to explore possible contributions from the field of Design to Education in the sense of bringing new perspectives to its role in students’ capacity of grasping the complexity of the world, as described by Morin. In order to accomplish that, a trajectory of epistemological and methodological discussions that question the congruence of the terms problem and solution with Design education and practice has been traced, from Simon to Coyne and Findeli, passing by Rittel AND Webber, Schön and Buchanan. Interviews with teachers who work in general Education and studied Design were also conducted as a way to engage ourselves in a dialogue from which concepts and connections about how Design thinking can contribute to the educational field could emerge. We understand that, because of its plural cultural dimension, the context of motivations and proclivities from Design favors the transcendence of dichotomies and the understanding of the fundamental aspect of complexity.

Student Number: 8402734 Master of Arts (Speech Pathology) Faculty of Humanities / Comprehension was assessed under various conditions in nine subjects with probable Alzheimer’s Disease. Four linguistic variables were selected, namely syntactic complexity, the use of pronouns, directness and literality. Each variable was assessed in sentences in a simple form and in a complex form. Discourse passages which included all of the variables were then constructed, and these tasks were also administered in simple and complex forms. All subjects performed qualitatively and quantitatively better under the simple conditions than under the complex for all four of the variables assessed. However, this effect was more marked in the sentence tasks than in the discourse tasks. Furthermore, individual subjects responded differently to the manipulation of the variables. Results of the study suggest that modifying the variables investigated can improve the comprehension of people with moderate to severe Alzheimer’s Disease. This has important implications for the training of caregivers and conversational partners, and suggests that this type of indirect intervention may be an effective way for speech therapists to provide clinical services to people with dementia in South Africa.

It has been long known that groups of adults learn and enact their learning in certain ways; what is little known is how groups learn and how they develop in cognitive complexity. This dissertation proposes a theory of group cognitive development by arguing that intentional adult groups are complex and dynamic, and that they have the potential to evolve over time. Groups are complex in that they are made up of individuals within different orders of consciousness (Kegan), and they are dynamic in that different orders of consciousness interact and conflict (Lonergan) during the formation and enactment of group vision, values, and procedures. Dynamic complexity theory of group development as it is referred to in this study is grounded in Robert Kegan’s constructive developmental theory and in Bernard Lonergan’s transcendental method. While both Kegan and Lonergan attend to the growth of individuals, their theories are adapted to groups in order to understand the cognitive complexity of groups, intragroup and intergroup conflict, and the mental complexity of leader curriculum. This theory is applied to two case studies, one from antiquity in the case of the first century Corinthian community engaged in conflict with its founder, St. Paul, and in one contemporary study of American Catholic parishioners engaged in contentious dialogue with diocesan leaders from 1994 to 2004. The parish groups experienced a series of dialogues during a ten year period over the issues of parish restructuring and the priest sexual abuse crisis yielding cumulative and progressive changes in perspective-taking, responsibility-taking, and in group capacity to respond to and engage local and institutional authority figures. Group development is observed against a pedagogical backdrop that represents a mismatch between group complexity and leader expectations. In Corinth, Paul’s curriculum was significantly beyond the mental capacity of the community. In the case of Catholic parishioners the curriculum of diocesan leaders was beneath the mental capacities of most of the groups studied. It is proposed that individuals sharing the same order of consciousness, understood as cognitive constituencies, are in a dynamic relationship with other cognitive constituencies in the group that interact within an object-subject dialectic and an agency-communion dialectic. The first describes and explains the evolving cognitive complexity of group knowing, how the group does its knowing, and what it knows when it is doing it (the epistemologies of the group). This dialectic has implications for how intentional groups might be the critical factor for understanding individual growth. The second dialectic describes and explains the changing relationship between group agency, which is enacted either instrumentally or ideologically; and group communion, which is enacted ideationally. The agency-communion dialectic is held in an unstable balance in the knowing, identity, and mission of groups. With implications for the fields of adult education and learning organizations, dynamic complexity theory of group development notes predictable stages of group evolution as each cognitive constituency evolves, and notes the significance of internal and external conflict for exposing the presence of different ways of knowing and for challenging the group toward cognitive growth. / Thesis (PhD) — Boston College, 2008. / Submitted to: Boston College. Graduate School of Arts and Sciences. / Discipline: Religious Education and Pastoral Ministry.

Cette thèse porte sur un certain nombre de problèmes algorithmiques motivés par la planification temporelle automatisée et la vérification formelle des systèmes réactifs et finis. Nous nous sommes concentrés sur les méthodes théoriques des jeux pour obtenir de nouvelles connaissances, des limites de complexité améliorées et des algorithmes plus rapides pour les modèles suivants: réseaux temporels hyper, réseaux conditionnels Simples / Hyper temporels, jeux de mise à jour, jeux Muller McNaughton et jeux Mean Payoff / This dissertation deals with a number of algorithmic problems motivated by automated temporal planning and formal verification of reactive and finite state systems. We focused on game theoretical methods to obtain novel insights, improved complexity bounds, and faster algorithms for the following models: Hyper Temporal Networks, Conditional Simple/Hyper Temporal Networks, Update Games, Muller McNaughton Games, and Mean Payoff Games

Esse estudo se inicia na filosofia subjacente à pesquisa constante da literatura dedicada à gestão de portfólios de projetos de desenvolvimento de novos produtos e termina na aplicabilidade das proposições para intervenção nesse sistema complexo. Contrapondo-se a realidade do mundo empresarial ao panorama de pesquisas acadêmicas sobre o tema, verificou-se que há uma lacuna entre o que é cientificamente proposto e a sua aplicabilidade no cotidiano empresarial. Portanto este estudo buscou analisar a complexidade da gestão de portfólio de projetos de desenvolvimento de novos produtos, procurando determinar os paradigmas adotados para seu estudo e para as suas proposições. Também se propôs uma metodologia analítica baseada na analogia ao mercado consumidor, sistema tão ou mais complexo que o portfólio de desenvolvimento de novos produtos. Para tanto, utilizou-se do método indutivo, partindo de um conjunto de conhecimentos observáveis na revisão sistemática de literatura, além da aplicação dessa nova metodologia analítica, com seu ferramental próprio, com alta carga de abstração. Foi então possível evidenciar a complexidade do objeto de pesquisa e a incongruência epistemológica e paradigmática da pesquisa sobre o tema, que, em grande medida, utiliza níveis de abstração diferentes para o objeto da pesquisa e para suas proposições, incorrendo com isso, em dificuldades na aplicabilidade de suas propostas. No longo prazo, essa metodologia pode vir a ser de grande utilidade, não só para auxiliar os gestores e profissionais em geral, que trabalham com desenvolvimento de produtos, como para melhorar a aplicabilidade das proposições acadêmicas no âmbito empresarial. / This study starts with the philosophy underlying the research literature dedicated to the management of new product development project portfolios and ends with the application of the proposals for intervention in that complex system. Contrasting the reality of the business world to the panorama of academic research on the subject, it was verified that there is a gap between what is scientifically proposed and its applicability in daily business. Therefore, the aim of this study is to analyze the complexity of the new product development project portfolios, trying to determine the paradigms adopted for its study and for its propositions. An analytical methodology based on consumer market, a system as complex as the new product development portfolio, was also proposed. For this much, the inductive method was utilized, starting from a set of observable knowledge in the systematic review of literature, in addition to the new methodology and its own tooling, with a high load of abstraction. It was then possible to highlight the complexity of the research object and an epistemological and paradigmatic incongruity on the research subject, which, to a large extent, ends up using different levels of abstraction for the research object and for its proposals. In the long term, the new methodology can be very useful, not only to help managers and professionals in general who work with product development, but also to improve the application of academic propositions in business.

[pt] O sistema jurídico kelseniano, através de uma releitura heterodoxa, pode ser resgatado como um modelo eficiente para o paradigma que emerge. Tal assertiva se baseia na convicção de que as críticas impostas a este autor muitas vezes são fruto da incompreensão de sua proposta, o que se dá por uma confusão de pressupostos. A análise metateórica é um método que permite recolocar o pensamento kelseniano não como algo ultrapassado, mas como um pensamento de vanguarda, apto a ser utilizado a serviço da Pós - Modernidade. Ainda dentro do arcabouço paradigmático Moderno, os pressupostos kelsenianos são associados a um referencial teórico que se afasta da concepção da linguagem como representação do real, explicitando assim , na Teoria Pura do Direito, uma crítica parcial à Modernidade. Kelsen, apesar de não ter tido a pretensão de se imiscuir em determinadas áreas por uma questão epistemológica, deixou em sua teoria espaços abertos que hoje, à luz de novos olhares trazidos especialmente pelo desenvolvimento de áreas como a psicanálise, podem ser usados para se fazer uma crítica ampla e não apenas parcial à Modernidade. A partir de uma análise metapsicológica das noções de liberdade e igualdade na teoria política de Kelsen e da atualização de sua concepção de natureza humana, abre- se uma nova perspectiva, através da qual é possível se fazer uma leitura mais apurada da Teoria Pura do Direito, onde a ética do sentido se explicita como condição de validade dos ordenamentos, a tornando um instrumental, não mais obsoleto, e sim adequado a enfrentar os problemas da contemporaneidade. / [en] Kelsen´s juridical system, through a heterodox perspective, could be seen as an efficient model for the emerging paradigm, based upon the belief that the criticism of this author is due to misunderstanding of his proposals and/or hypothesis. The metatheoretical analysis is one of the methods that reconduct Kelsen`s thoughts not as something antique, but as a breaktrough that could be used at the service of Post-Modernity. Still within the Modern paradigm, Kelsen´s hypothesis are associated with a theoretical referential that is away of the language`s conception as a representation of the real, highlighting, in Law´s Pure Theory, a partial critique to Modernity. Kelsen, even not having the proposal of asserting at some areas due to epistemic reasons, left blank spaces in his theory that, using some new insights of areas such as psychoanalysis, could be used as tools of a complete, not partial, critique to Modernity. Using a metapsychological analysis of Kelsen`s political theory concepts of liberty and equality and updating his concept of human nature, there is a new perspective through which is possible to have a proper reading of Law´s Pure Theory, where sense`s ethics is an explicit condition to validation of juridical ordering and tool, non obsolete, to deal with the problems of contemporaniety.

Hållbar utveckling är ett omtalat och ständigt närvarande begrepp i dagens samhälle likväl som i läroplanen för grundskolan (Skolverket 2018a). Begreppet hållbar utveckling nämns återkommande i läroplanen. Att undervisning för hållbar utveckling är avgörande för att eleverna ska utveckla kunskap, förståelse och viljan att förändra lyfts av Skolverket (2018a). Läraren har således ett stort ansvar för att genom samverkan med eleverna utveckla undervisning som uppmuntrar till hållbar utveckling. Då läraren besitter ett stort ansvar kring att inkludera hållbar utveckling i undervisningen är det i högsta grad relevant att undersöka lärarens syn på hållbar utveckling och dess roll i undervisningen. Studiens syfte är att belysa lärares beskrivningar om hållbar utveckling och att få en inblick i hur hållbar utveckling kommer till uttryck i undervisningen med utgångspunkt i ämnet biologi. Utifrån studiens syfte genomförs en kvalitativ undersökning i form av semistrukturerade intervjuer, där åtta lärare från två kommuner i Mellansverige deltar. Studiens resultat visar på en variation av beskrivningar, där begreppets komplexitet lyfts. Två typer av förhållningssätt kring hållbar utveckling framkommer. Det ena förhållningssättet innefattar sporadisk undervisning med ett stort miljöfokus medan det andra präglas av ett inkluderande arbetssätt, där ett hållbart perspektiv finns närvarande i all undervisning. Resultatet poängterar betydelsen av diskussioner och samtal för att främja elevers handlingsförmåga. Fortbildning på området lyser med sin frånvaro samtidigt som svårigheten med begreppets bredd och komplexitet poängteras. / Sustainable development is a well-known and ever-present concept in today's society as well as in the curriculum for compulsory school (Skolverket 2018a). The concept of sustainable development is mentioned repeatedly in the curriculum. Education for sustainable development is crucial for the students in order to develop knowledge, understanding and the desire to change (Skolverket 2018a). The teacher thus has a great responsibility to, through collaboration with the students, develop teaching that encourages sustainable development. Since teachers have a great responsibility for promoting sustainable development in their teaching, it is highly relevant to examine the teacher's view of sustainable development and its role in teaching. The purpose of the study is to highlight teachers' descriptions of sustainable development and to achieve an awareness of how sustainable development is expressed in teaching based on the subject of biology. Based on the study's purpose, a qualitative study is carried out in the form of semi-structured interviews, where eight teachers from two municipalities in Central Sweden participated. The results show a variety of descriptions, where the complexity of the concept is pointed out. In the analysis, two types of approaches to sustainable development were identified. One approach includes sporadic teaching with a focus on environmental aspects while the other is characterized by an inclusive approach in teaching, where sustainable development is integrated in all teaching. The results indicate the importance of discussions and conversations to promote the ability to act. In-service training in the field shines with its absence, while the teachers emphasize the difficulty with extent and complexity of the concept.

Stream restoration efforts often aim at restoring the physical complexity in streams, as an increased habitat heterogeneity is believed to increase biodiversity. It is important to quantify the physical complexity of streams before and after restoration, to know what actions are needed, and to monitor the results of the restoration. The use of unmanned aerial vehicles (UAVs) and geographic information systems (GIS) for data acquisition is rapidly increasing, and the use of UAVs and GIS could facilitate the monitoring process. The aim of this study was to determine how the spatial complexity in streams can be determined by using UAVs and GIS. The physical features and the spatial complexity were quantified in five reaches in the Lögde River, pre- and post-restoration, by analyzing UAV photos in a GIS program. Three of six reach descriptive metrics, and three of seven complexity metrics, were shown significantly different after restoration. To validate the GIS analyzing method, a qualitative comparison of data from the GIS analysis to field survey data was conducted. The GIS method was shown effective for distinguishing morphological features on a larger spatial scale, and to show the spatial distribution of instream features, such as wood pieces and boulders. The accuracy when digitizing the bankfull edge of the stream was low on small scales, and the method likely underestimates the number of wood pieces and boulders in the streams. Preferable camera settings and weather conditions to avoid blurry UAV photos, and thereby enhance the accuracy of the GIS analysis, are discussed.

Nous considérons un problème d'ordonnancement à une machine avec dates de fin impératives et nous cherchons caractériser l'ensemble des solutions optimales, sans les énumérer. Nous supposons que les travaux sont numérotés selon la règle EDD et que cette séquence est réalisable. La méthode consiste à utiliser le treillis des permutations et d'associer à la permutation maximale du treillis la séquence EDD. Afin de caractériser beaucoup de solutions, nous cherchons une séquence réalisable aussi loin que possible de cette séquence. La distance utilisée est le niveau de la séquence dans le treillis, qui doit être minimum (le plus bas possible). Cette nouvelle fonction objectif est étudiée. Quelques cas particuliers polynomiaux sont identifiés, mais la complexité du problème général reste ouverte. Quelques méthodes de résolution, polynomiales et exponentielles, sont proposées et évaluées. Le niveau de la séquence étant en rapport avec la position des travaux dans la séquence, de nouvelles fonctions objectifs en rapport avec les positions des travaux sont identifiées et étudiées. Le problème de la minimisation de la somme pondérée des positions des travaux est prouvé fortement NP-difficile. Quelques cas particuliers sont étudiés et des méthodes de résolution proposées et évaluées. / We consider a single machine scheduling problem with deadlines and we want to characterise the set of optimal solutions, without enumerating them. We assume that jobs are numbered in EDD order and that this sequence is feasible. The key idea is to use the lattice of permutations and to associate to the supremum permutation the EDD sequence. In order to characterize a lot of solutions, we search for a feasible sequence, as far as possible to the supremum. The distance is the level of the sequence in the lattice, which has to be minimum. This new objective function is investigated. Some polynomially particular cases are identified, but the complexity of the general case problem remains open. Some resolution methods, polynomial and exponential, are proposed and evaluated. The level of the sequence being related to the positions of jobs in the sequence, new objective functions related to the jobs positions are identified and studied. The problem of minimizing the total weighted positions of jobs is proved to be strongly NP-hard. Some particular cases are investigated, resolution methods are also proposed and evaluated.