• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 2
  • 2
  • 1
  • Tagged with
  • 5
  • 5
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 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.
1

Model-based clustering based on sparse finite Gaussian mixtures

Malsiner-Walli, Gertraud, Frühwirth-Schnatter, Sylvia, Grün, Bettina January 2016 (has links) (PDF)
In the framework of Bayesian model-based clustering based on a finite mixture of Gaussian distributions, we present a joint approach to estimate the number of mixture components and identify cluster-relevant variables simultaneously as well as to obtain an identified model. Our approach consists in specifying sparse hierarchical priors on the mixture weights and component means. In a deliberately overfitting mixture model the sparse prior on the weights empties superfluous components during MCMC. A straightforward estimator for the true number of components is given by the most frequent number of non-empty components visited during MCMC sampling. Specifying a shrinkage prior, namely the normal gamma prior, on the component means leads to improved parameter estimates as well as identification of cluster-relevant variables. After estimating the mixture model using MCMC methods based on data augmentation and Gibbs sampling, an identified model is obtained by relabeling the MCMC output in the point process representation of the draws. This is performed using K-centroids cluster analysis based on the Mahalanobis distance. We evaluate our proposed strategy in a simulation setup with artificial data and by applying it to benchmark data sets. (authors' abstract)
2

Sparse Optimal Control for Continuous-Time Dynamical Systems / 連続時間システムに対するスパース最適制御

Ikeda, Takuya 25 March 2019 (has links)
京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第21916号 / 情博第699号 / 新制||情||120(附属図書館) / 京都大学大学院情報学研究科数理工学専攻 / (主査)准教授 加嶋 健司, 教授 太田 快人, 教授 山下 信雄 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM
3

[pt] LAWIE: DECONVOLUÇÃO EM PICOS ESPARSOS USANDO O LASSO E FILTRO DE WIENER / [en] LAWIE: SPARSE-SPIKE DECONVOLUTION WITH LASSO AND WIENER FILTER

FELIPE JORDAO PINHEIRO DE ANDRADE 06 November 2020 (has links)
[pt] Este trabalho propõe um algoritmo para o problema da deconvolução sísmica em picos esparsos. Intitulado LaWie, este algoritmo é baseado na combinação do Least Absolute Shrinkage and Selection Operator (LASSO) e a modelagem de blocos usada no filtro de Wiener. A deconvolução é feita traço a traço para estimar o perfil de refletividade e a wavelet original que deu origem as amplitudes sísmicas. Este trabalho apresenta o resultado do método no dataset sintético do Marmousi2, onde existe um ground truth para comparações objetivas. Além disso, também apresenta os resultados no dataset real Netherlands Offshore F3 Block e mostra a aplicabilidade do algoritmo proposto para não apenas delinear o perfil de refletividades como também para ressaltar características como fraturas neste dado. / [en] This work proposes an algorithm for solving the seismic sparse-spike deconvolution problem. Entitled LaWie, this algorithm is based on the combination of Least Absolute Shrinkage and Selection Operator (LASSO) and the block modeling used in the Wiener filter. Deconvolution is done trace by trace to estimate the reflectivity profile and the convolution wavelet that originated the seismic amplitudes. This work presents the results in the synthetic dataset of Marmousi2, where there is a ground truth for objective comparisons. Also, this work presents the results in a real dataset, Netherlands Offshore F3 Block, and shows the applicability of the proposed algorithm to outline the reflectivity profile and highlight characteristics such as fractures in this data.
4

[pt] MODELAGEM ESPARSA E SUPERTRAÇOS PARA DECONVOLUÇÃO E INVERSÃO SÍSMICAS / [en] SPARSE MODELING AND SUPERTRACES FOR SEISMIC DECONVOLUTION AND INVERSION

RODRIGO COSTA FERNANDES 11 May 2020 (has links)
[pt] Dados de amplitude sísmica compõem o conjunto de insumos do trabalho de interpretação geofísica. À medida que a qualidade dos sensores sísmicos evoluem, há aumento importante tanto na resolução quanto no espaço ocupado para armazenamento. Neste contexto, as tarefas de deconvolução e inversão sísmicas se tornam mais custosas, em tempo de processamento ou em espaço ocupado, em memória principal ou secundária. Partindo do pressuposto de que é possível assumir, por aproximação, que traços de amplitudes sísmicas são o resultado da fusão entre um conteúdo oscilatório – um pulso gerado por um tipo de explosão, em caso de aquisição marítima – e a presença esparsa de contrastes de impedância e variação de densidade de rocha, pretende-se, neste trabalho, apresentar contribuições quanto à forma de realização de duas atividades em interpretação geofísica: a deconvolução e a inversão de refletividades em picos esparsos. Tomando como inspiração trabalhos em compressão volumétrica 3D e 4D, modelagem esparsa, otimização em geofísica, segmentação de imagens e visualização científica, apresenta-se, nesta tese, um conjunto de métodos que buscam estruturas fundamentais e geradoras das amplitudes: (i) uma abordagem para segmentação e seleção de traços sísmicos como representantes de todo o dado, (ii) uma abordagem para separação de amplitudes em ondaleta e picos esparsos de refletividade via deconvolução e (iii) uma outra para confecção de um operador linear – um dicionário – capaz de representar, parcial e aproximadamente, variações no conteúdo oscilatório – emulando alguns efeitos do subsolo –, com o qual é possível realizar uma inversão de refletividades. Por fim, apresentase um conjunto de resultados demonstrando a viabilidade das abordagens, o ganho eventual se aplicadas – incluindo a possibilidade de compressão – e a abertura de oportunidades de trabalhos futuros mesclando geofísica e computação. / [en] Seismic amplitude data are part of the input in a geophysical interpretation pipeline. As seismic sensors evolve, resolution and occupied storage space grows. In this context, tasks as seismic deconvolution and inversion become more expensive, in processing time or in – main or secondary – memory. Assuming that, approximately, seismic amplitude traces result from a fusion between an oscillatory content – a pulse generated by a kind of explosion, in the case of marine acquisition – and the sparse presence of impedance constrasts and rock density variation, this work presents contributions to the way of doing two geophysical interpretation activities: deconvolution and inversion, both targeting sparse-spike refletivity extraction. Inspired by works in 3D and 4D volumetric compression, sparse modeling, optimization applied to geophysics, image segmentation and scientific visualization, this thesis presents a set of methods that try to fetch fundamental features that generate amplitude data: (i) an approach for seismic traces segmentation and selection, electing them as representatives of the whole data, (ii) an enhancement of an approach for separation of amplitudes into wavelet and sparse-spike reflectivities via deconvolution, and (iii) a way to generate a linear operator – a dictionary – partially and approximately capable of representing variations on the wavelet shape, emulating some effects of the subsoil, from which is possible to accomplish a reflectivity inversion. By the end, it is presented a set of results that demonstrate the viability of such approaches, the possible gain when they are applied – including compression – and some opportunities for future works mixing geophysics and computer science.
5

Parcimonie, diversité morphologique et séparation robuste de sources / Sparse modeling, morphological diversity and robust source separation

Chenot, Cécile 29 September 2017 (has links)
Cette thèse porte sur le problème de Séparation Aveugle de Sources (SAS) en présence de données aberrantes. La plupart des méthodes de SAS sont faussées par la présence de déviations structurées par rapport au modèle de mélange linéaire classique: des évènements physiques inattendus ou des dysfonctionnements de capteurs en sont des exemples fréquents.Nous proposons un nouveau modèle prenant en compte explicitement les données aberrantes. Le problème de séparation en résultant, mal posé, est adressé grâce à la parcimonie. L'utilisation de cette dernière est particulièrement intéressante en SAS robuste car elle permet simultanément de démélanger les sources et de séparer les différentes contributions. Ces travaux sont étendus pour l'estimation de variabilité spectrale pour l'imagerie hyperspectrale terrestre.Des comparaisons avec des méthodes de l'état-de-l'art montrent la robustesse et la fiabilité des algorithmes associés pour un large éventail de configurations, incluant le cas déterminé. / This manuscript addresses the Blind Source Separation (BSS) problem in the presence of outliers. Most BSS techniques are hampered by the presence of structured deviations from the standard linear mixing model, such as unexpected physical events or malfunctions of sensors. We propose a new data model taking explicitly into account the deviations. The resulting joint estimation of the components is an ill-posed problem, tackled using sparse modeling. The latter is particularly efficient for solving robust BSS since it allows for a robust unmixing of the sources jointly with a precise separation of the components. These works are then extended for the estimation of spectral variability in the framework of terrestrial hyperspectral imaging. Numerical experiments highlight the robustness and reliability of the proposed algorithms in a wide range of settings, including the full-rank regime.

Page generated in 0.088 seconds