• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 41
  • 5
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • Tagged with
  • 70
  • 22
  • 21
  • 14
  • 13
  • 11
  • 10
  • 10
  • 10
  • 9
  • 9
  • 9
  • 8
  • 8
  • 7
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
21

Julia Wolfe’s Zigzag for Wind Band: The Commission, Analysis, and Interviews

Farr, Daniel Kenneth 09 August 2022 (has links)
No description available.
22

The relation between the poetic concept and autobiographical memory in the works of Thomas Wolfe

Taylor, Douglas Carell 01 January 1949 (has links) (PDF)
Thomas Wolfe's literary output includes four long novels and a vast number of short stories and poems. In addition to this he left a great deal of manuscript which is still in the process of being examined and assorted. His four long novels have drawn the greatest attention, for it is in these that he has done his finest and most provocative work. In each one Wolfe is the central figure, and through his eyes we are allowed to see the world as he saw it. It is because of this that the charges of egotism, paranoia tendencies, genius, immaturity, and plain mediocrity have been hurled at him. From the statement of Sinclair Lewis that Wolfe might live to be the greatest writer that America has yet produced, to Canby's "I think that this novel Of Time and the Riverlike many fiery and ambitious American books ... is an artistic flop," Wolfe has been analyzed and interpreted by critics, near critics, and uncritical sentiment. The most consistent criticism leveled at him is that whatever success he may have attained is due to autobiographical memory, and that if he laid aside the tools of subjectivity, his work would have only an ordinary prosaic quality. It is the purpose of this study to demonstrate that Wolfe's accomplishments were due not to autobiographical memory but instead to a supreme intensity of poetic feeling and tremendous scope of power and imagery--to show that Thomas Wolfe was an artist who was abundantly rich in word range, a master of characterization and of creative imagination, thereby to refute the claim of those who narrow his potentialities merely to those of an autobiographical nature.
23

Stochastic Frank-Wolfe Algorithm : Uniform Sampling Without Replacement

Håkman, Olof January 2023 (has links)
The Frank-Wolfe (FW) optimization algorithm, due to its projection free property, has gained popularity in recent years with typical application within the field of machine learning. In the stochastic setting, it is still relatively understudied in comparison to the more expensive projected method of Stochastic Gradient Descent (SGD). We propose a novel Stochastic Frank-Wolfe (SFW) algorithm, inspired by SGDo where random sampling is considered without replacement. In analogy to this abbreviation, we call the proposed algorithm SFWo. We consider a convex setting of gradient Lipschitz (smooth) functions over compact domains. Depending on experiment design (LASSO, Matrix Sensing, Matrix Completion), the SFWo algorithm, exhibits a faster or matching empirical convergence, and a tendency of bounded suboptimality by the precursor SFW. Benchmarks on both synthetic and real world data display that SFWo improves on the number of stochastic gradient evaluations needed to achieve the same guarantee as SFW. / Intresset för Frank-Wolfes (FW) optimeringsalgoritm har tack vare dess projektionsfria egenskap ökat de senaste åren med typisk tillämpning inom området maskininlärning. I sitt stokastiska uförande är den fortfarande relativt understuderad i jämförelse med den dyrare projicerande metoden Stochastic Gradient Descent (SGD). Vi föreslår en ny Stochastic Frank-Wolfe(SFW) algoritm, inspirerad av SGDo där slumpmässigt urval görs utan återläggning. I analogi med denna förkortning så kallar vi den föreslagna algoritmen SFWo. Vi betraktar en konvex miljö och gradient Lipschitz kontinuerliga (släta) funktioner över kompakta definitionsmängder. Beroende på experimentdesign (LASSO, Matrix Sensing, Matrix Completion) så visar den föreslagna algoritmen SFWo på en snabbare eller matchande empirisk konvergens och tenderar vara begränsad i suboptimalitet av föregångaren SFW. Prestandajämförelser på både syntetisk och verklig data visar att SFWo förbättrarantalet stokastiska gradientevalueringar som behövs för att uppnå samma garanti som för SFW.
24

Résolution exacte de problèmes de couverture par arborescences sous contraintes de capacité / Exact methods for solving covering problems with trees subject to capacity constraints

Guillot, Jérémy 18 December 2018 (has links)
Dans ce document, nous étudions deux problèmes de sectorisation et proposons plusieurs méthodes de résolution exactes basées sur la décomposition de Dantzig-Wolfe et la génération de colonnes. Nous proposons deux modélisations en fonction de la manière d’appréhender l’objectif du problème qui consiste à obtenir des secteurs compacts. Pour chacune des modélisations, nous comparons des approches de résolution exactes basées sur des formulations compactes ou sur des formulations étendues obtenues par la décomposition de Dantzig-Wolfe. Le premier type de modèles proposé définit la fonction objectif à la manière d’un problème de p-median. Concernant les méthodes de résolution pour ce type de modèle, l’accent est mis sur l’accélération de la convergence de l’algorithme de génération de colonnes en mettant en place des techniques d’agrégation de contraintes afin de réduire la dégénérescence de l’algorithme du simplexe. Les expérimentations numériques montrent que la méthode d’agrégation de contraintes proposée permet effectivement de réduire le nombre d’itérations dégénérées. Cependant, elle ne suffit pas à accélérer l’algorithme de branch-and-price. Le choix d’utilisation de la formulation compacte ou de la formulation étendue dépend du type d’instances résolu. Le second type de modèles formule l’objectif d’une manière assez proche de celui des problèmes de p-centre. L’utilisation d’un tel objectif complexifie la résolution des sous-problèmes de génération de colonnes. L’accent est donc mis sur la conception d’algorithmes de branch-and-bound et de programmation dynamique pour les résoudre efficacement. Les expériences montrent que l’algorithme de branch-and-price surpasse les approches de résolution utilisant une formulation compacte du problème. / In this document, we study two districting problems and propose several exact methods, based on Dantzig-Wolfe decomposition and column generation, to solve them. For each model, we compare exact approaches based either on compact formulations or on extended formulations obtained using Dantzig-Wolfe decomposition. The first type of model that we propose defines the objective function in a p-median problem fashion. Regarding the methods used to solve that kind of model, we emphasize accelerating the convergence of the column generation algorithm by designing constraint aggregation techniques in order to reduce the degeneracy in the simplex algorithm. Numerical experiments show that this constraint aggregation method indeed reduces the proportion of degenerated iterations. However, it is not enough to speed up the branch-and-price algorithm. Choosing to tackle the problem through either a compact formulation or an extended formulation depends on the structure of the instances to solve. The second type of model formulates the objective function in a way quite similar to that of p-centre problems. Using such an objective function induces complex column generation subproblems. We focus on designing branch-and-bound and dynamic programming algorithms in order to solve them efficiently. Experiments show that the branch-and-price approach surpasses any proposed method based on compact formulations of the problem.
25

Geração de colunas para o problema de dimensionamento de lotes de produção com limitações de capacidade / Column generation heuristics for capacitated lotsizing problem

Baldo, Tamara Angélica 29 May 2009 (has links)
O problema de dimensionamento de lotes com restrições de capacidade (CLSP) consiste em determinar um plano de produção que satisfaça a demanda requerida, respeitando as limitações de capacidade, com o menor custo possível, ou seja, minimizando os custos de produção, estocagem e preparação de máquina. Encontrar uma solução factível para o CLSP, considerando tempo de preparação de máquina, é NP-completo. Nesta dissertação, para a resolução do CLSP, utiliza-se a decomposição de Dantzig-Wolfe e o procedimento de geração de colunas, encontrando bons limitantes inferiores. Duas diferentes estratégias de decomposição são exploradas, decomposição por itens e períodos. Para a obtenção de uma solução inteira para o problema (limitante superior) foram exploradas heurísticas lagrangianas, onde a solução inicial para as heurísticas provém da geração de colunas. Os limitantes obtidos podem ser utilizados em métodos exatos, como por exemplo, em algoritmos do tipo branch-and-price. Experimentos computacionais, baseados em exemplares gerados aleatoriamente, foram realizados e os resultados analisados, as variações dos parâmetros das instâncias foram sugeridas na literatura / The Capacitated Lot Sizing Problem (CLSP) consists in determining a production plan such that all demands are met and the total costs of production, inventory and setup are minimized. Since the problem to find a feasible solution to the CLSP with setup times is NP-complete, large problem instances have been solved by heuristic methods. In this dissertation, we are particularly concerned in using the methodology of Dantzig-Wolfe decomposition and column generation to generate good bounds to the CLSP with setup times and costs. Here, we analyse two types of decomposition which are based on items and time periods (lower bound) and some lagrangian-based heuristics (upper bound). Numerical results based on randomly generated intances suggest that highquality lower bounds are obtained by column generation algorithms, such as well as upper bounds by heuristics. These bounds are useful in exact solution methods, such as branch-and-price algorithms
26

Optimisation non-lisse pour l'apprentissage statistique avec régularisation matricielle structurée / Nonsmooth optimization for statistical learning with structured matrix regularization

Pierucci, Federico 23 June 2017 (has links)
La phase d’apprentissage des méthodes d’apprentissage statistique automatique correspondent à la résolution d’un problème d’optimisation mathématique dont la fonction objectif se décompose en deux parties: a) le risque empirique, construit à partir d’une fonction de perte, dont la forme est déterminée par la métrique de performance et les hypothèses sur le bruit; b) la pénalité de régularisation, construite a partir d’une norme ou fonction jauge, dont la structure est déterminée par l’information à priori disponible sur le problème a résoudre.Les fonctions de perte usuelles, comme la fonction de perte charnière pour la classification supervisée binaire, ainsi que les fonctions de perte plus avancées comme celle pour la classification supervisée avec possibilité d’abstention, sont non-différentiables. Les pénalités de régularisation comme la norme l1 (vectorielle), ainsi que la norme nucléaire (matricielle), sont également non- différentiables. Cependant, les algorithmes d’optimisation numériques les plus simples, comme l’algorithme de sous-gradient ou les méthodes de faisceaux, ne tirent pas profit de la structure composite de l’objectif. Le but de cette thèse est d’étudier les problèmes d’apprentissage doublement non-différentiables (perte non- différentiable et régularisation non-différentiable), ainsi que les algorithmes d’optimisation numérique qui sont en mesure de bénéficier de cette structure composite.Dans le premier chapitre, nous présentons une nouvelle famille de pénalité de régularisation, les normes de Schatten par blocs, qui généralisent les normes de Schatten classiques. Nous démontrons les principales propriétés des normes de Schatten par blocs en faisant appel à des outils d’analyse convexe et d’algèbre linéaire; nous retrouvons en particulier des propriétés caractérisant les normes proposées en termes d’enveloppe convexes. Nous discutons plusieurs applications potentielles de la norme nucléaire par blocs, pour le filtrage collaboratif, la compression de bases de données, et l’annotation multi-étiquettes d’images.Dans le deuxième chapitre, nous présentons une synthèse de différentes tech- niques de lissage qui permettent d’utiliser des algorithmes de premier ordre adaptes aux objectifs composites qui de décomposent en un terme différentiable et un terme non-différentiable. Nous montrons comment le lissage peut être utilisé pour lisser la fonction de perte correspondant à la précision au rang k, populaire pour le classement et la classification supervises d’images. Nous décrivons dans les grandes lignes plusieurs familles d’algorithmes de premier ordre qui peuvent bénéficier du lissage: i) les algorithmes de gradient conditionnel; ii) les algorithmes de gradient proximal; iii) les algorithmes de gradient incrémental.Dans le troisième chapitre, nous étudions en profondeur les algorithmes de gradient conditionnel pour les problèmes d’optimisation non-différentiables d’apprentissage statistique automatique. Nous montrons qu’une stratégie de lis- sage adaptative associée à un algorithme de gradient conditionnel donne lieu à de nouveaux algorithmes de gradient conditionnel qui satisfont des garanties de convergence théoriques. Nous présentons des résultats expérimentaux prometteurs des problèmes de filtrage collaboratif pour la recommandation de films et de catégorisation d’images. / Training machine learning methods boils down to solving optimization problems whose objective functions often decomposes into two parts: a) the empirical risk, built upon the loss function, whose shape is determined by the performance metric and the noise assumptions; b) the regularization penalty, built upon a norm, or a gauge function, whose structure is determined by the prior information available for the problem at hand.Common loss functions, such as the hinge loss for binary classification, or more advanced loss functions, such as the one arising in classification with reject option, are non-smooth. Sparse regularization penalties such as the (vector) l1- penalty, or the (matrix) nuclear-norm penalty, are also non-smooth. However, basic non-smooth optimization algorithms, such as subgradient optimization or bundle-type methods, do not leverage the composite structure of the objective. The goal of this thesis is to study doubly non-smooth learning problems (with non-smooth loss functions and non-smooth regularization penalties) and first- order optimization algorithms that leverage composite structure of non-smooth objectives.In the first chapter, we introduce new regularization penalties, called the group Schatten norms, to generalize the standard Schatten norms to block- structured matrices. We establish the main properties of the group Schatten norms using tools from convex analysis and linear algebra; we retrieve in particular some convex envelope properties. We discuss several potential applications of the group nuclear-norm, in collaborative filtering, database compression, multi-label image tagging.In the second chapter, we present a survey of smoothing techniques that allow us to use first-order optimization algorithms designed for composite objectives decomposing into a smooth part and a non-smooth part. We also show how smoothing can be used on the loss function corresponding to the top-k accuracy, used for ranking and multi-class classification problems. We outline some first-order algorithms that can be used in combination with the smoothing technique: i) conditional gradient algorithms; ii) proximal gradient algorithms; iii) incremental gradient algorithms.In the third chapter, we study further conditional gradient algorithms for solving doubly non-smooth optimization problems. We show that an adaptive smoothing combined with the standard conditional gradient algorithm gives birth to new conditional gradient algorithms having the expected theoretical convergence guarantees. We present promising experimental results in collaborative filtering for movie recommendation and image categorization.
27

Geração de colunas para o problema de dimensionamento de lotes de produção com limitações de capacidade / Column generation heuristics for capacitated lotsizing problem

Tamara Angélica Baldo 29 May 2009 (has links)
O problema de dimensionamento de lotes com restrições de capacidade (CLSP) consiste em determinar um plano de produção que satisfaça a demanda requerida, respeitando as limitações de capacidade, com o menor custo possível, ou seja, minimizando os custos de produção, estocagem e preparação de máquina. Encontrar uma solução factível para o CLSP, considerando tempo de preparação de máquina, é NP-completo. Nesta dissertação, para a resolução do CLSP, utiliza-se a decomposição de Dantzig-Wolfe e o procedimento de geração de colunas, encontrando bons limitantes inferiores. Duas diferentes estratégias de decomposição são exploradas, decomposição por itens e períodos. Para a obtenção de uma solução inteira para o problema (limitante superior) foram exploradas heurísticas lagrangianas, onde a solução inicial para as heurísticas provém da geração de colunas. Os limitantes obtidos podem ser utilizados em métodos exatos, como por exemplo, em algoritmos do tipo branch-and-price. Experimentos computacionais, baseados em exemplares gerados aleatoriamente, foram realizados e os resultados analisados, as variações dos parâmetros das instâncias foram sugeridas na literatura / The Capacitated Lot Sizing Problem (CLSP) consists in determining a production plan such that all demands are met and the total costs of production, inventory and setup are minimized. Since the problem to find a feasible solution to the CLSP with setup times is NP-complete, large problem instances have been solved by heuristic methods. In this dissertation, we are particularly concerned in using the methodology of Dantzig-Wolfe decomposition and column generation to generate good bounds to the CLSP with setup times and costs. Here, we analyse two types of decomposition which are based on items and time periods (lower bound) and some lagrangian-based heuristics (upper bound). Numerical results based on randomly generated intances suggest that highquality lower bounds are obtained by column generation algorithms, such as well as upper bounds by heuristics. These bounds are useful in exact solution methods, such as branch-and-price algorithms
28

A distributed Frank-Wolfe framework for trace norm minimization via the bulk synchronous parallel model / Une structure Frank-Wolfe distribuée pour la minimisation des normes de trace via le modèle parallèle synchrone en bloc

Zheng, Wenjie 13 June 2018 (has links)
L'apprentissage des matrices de rang faible est un problème de grande importance dans les statistiques, l'apprentissage automatique, la vision par ordinateur et les systèmes de recommandation. En raison de sa nature NP-difficile, une des approches principales consiste à résoudre sa relaxation convexe la plus étroite : la minimisation de la norme de trace. Parmi les différents algorithmes capables de résoudre cette optimisation, on peut citer la méthode de Frank-Wolfe, particulièrement adaptée aux matrices de grande dimension. En préparation à l'utilisation d'infrastructures distribuées pour accélérer le calcul, cette étude vise à explorer la possibilité d'exécuter l'algorithme de Frank-Wolfe dans un réseau en étoile avec le modèle BSP (Bulk Synchronous Parallel) et à étudier son efficacité théorique et empirique. Concernant l'aspect théorique, cette étude revisite le taux de convergence déterministe de Frank-Wolfe et l'étend à des cas non déterministes. En particulier, il montre qu'avec le sous-problème linéaire résolu de manière appropriée, Frank-Wolfe peut atteindre un taux de convergence sous-linéaire à la fois en espérance et avec une probabilité élevée. Cette contribution pose la fondation théorique de l'utilisation de la méthode de la puissance itérée ou de l'algorithme de Lanczos pour résoudre le sous-problème linéaire de Frank-Wolfe associé à la minimisation de la norme de trace. Concernant l'aspect algorithmique, dans le cadre de BSP, cette étude propose et analyse quatre stratégies pour le sous-problème linéaire ainsi que des méthodes pour la recherche linéaire. En outre, remarquant la propriété de mise à jour de rang-1 de Frank-Wolfe, il met à jour le gradient de manière récursive, avec une représentation dense ou de rang faible, au lieu de le recalculer de manière répétée à partir de zéro. Toutes ces conceptions sont génériques et s'appliquent à toutes les infrastructures distribuées compatibles avec le modèle BSP. Concernant l'aspect empirique, cette étude teste les conceptions algorithmiques proposées dans un cluster Apache SPARK. Selon les résultats des expériences, pour le sous-problème linéaire, la centralisation des gradients ou la moyenne des vecteurs singuliers est suffisante dans le cas de faible dimension, alors que la méthode de la puissance itérée distribuée, avec aussi peu qu'une ou deux itérations par époque, excelle dans le cas de grande dimension. La librairie Python développée pour les expériences est modulaire, extensible et prête à être déployée dans un contexte industriel. Cette étude a rempli sa fonction de preuve de concept. Suivant le chemin qu'il met en place, des solveurs peuvent être implémentés pour différentes infrastructures, parmi lesquelles des clusters GPU, pour résoudre des problèmes pratiques dans des contextes spécifiques. En outre, ses excellentes performances dans le jeu de données ImageNet le rendent prometteur pour l'apprentissage en profondeur. / Learning low-rank matrices is a problem of great importance in statistics, machine learning, computer vision, recommender systems, etc. Because of its NP-hard nature, a principled approach is to solve its tightest convex relaxation : trace norm minimization. Among various algorithms capable of solving this optimization is the Frank-Wolfe method, which is particularly suitable for high-dimensional matrices. In preparation for the usage of distributed infrastructures to further accelerate the computation, this study aims at exploring the possibility of executing the Frank-Wolfe algorithm in a star network with the Bulk Synchronous Parallel (BSP) model and investigating its efficiency both theoretically and empirically. In the theoretical aspect, this study revisits Frank-Wolfe's fundamental deterministic sublinear convergence rate and extends it to nondeterministic cases. In particular, it shows that with the linear subproblem appropriately solved, Frank-Wolfe can achieve a sublinear convergence rate both in expectation and with high probability. This contribution lays the theoretical foundation of using power iteration or Lanczos iteration to solve the linear subproblem for trace norm minimization. In the algorithmic aspect, within the BSP model, this study proposes and analyzes four strategies for the linear subproblem as well as methods for the line search. Moreover, noticing Frank-Wolfe's rank-1 update property, it updates the gradient recursively, with either a dense or a low-rank representation, instead of repeatedly recalculating it from scratch. All of these designs are generic and apply to any distributed infrastructures compatible with the BSP model. In the empirical aspect, this study tests the proposed algorithmic designs in an Apache SPARK cluster. According to the experiment results, for the linear subproblem, centralizing the gradient or averaging the singular vectors is sufficient in the low-dimensional case, whereas distributed power iteration, with as few as one or two iterations per epoch, excels in the high-dimensional case. The Python package developed for the experiments is modular, extensible and ready to deploy in an industrial context. This study has achieved its function as proof of concept. Following the path it sets up, solvers can be implemented for various infrastructures, among which GPU clusters, to solve practical problems in specific contexts. Besides, its excellent performance in the ImageNet dataset makes it promising for deep learning.
29

Creating a Sense of Relevancy: Directing Lee Blessing’s Two Rooms

Wolfe, Erick L 13 May 2016 (has links)
The following thesis is a documentation of the production of Two Rooms, including analysis, research production book, script, documentation, and evaluation. The play was performed in New Orleans, Louisiana, at the University of New Orleans Performing Arts Center Lab Theatre on November 7th, 8th, 12th, 13th, 21st, and 22nd, 2015.
30

Equipping a volunteer group at First Baptist Church, Wolfe City, Texas, to develop "personal crisis testimonies" which express thanks to God for his faithfulness in providing strength for recovering from difficulties so that participants will use their experiences in comforting others who are hurting and witnessing to the lost /

Mitchell, Kevin G. January 2005 (has links)
Thesis (D. Min.)--Midwestern Baptist Theological Seminary, 2005. / Abstract. Includes bibliographical references (leaves 275-281)

Page generated in 0.0224 seconds