102,616 research outputs found
The innovation of population forecasting methodology in the inter-war period: The case of the Netherlands
No full textLetter, [Author unclear] to Paulina T. Merritt
No full textHandwritten letter to Paulina Merritt from an unknown author, October 1, 1876.
Flow beneath inland navigation vessels
No full textGrowing transportation rates and the subsequent growth in inland waterway transport have led to an increase in inland vessel sizes and draught. Due to the fluctuating water levels on rivers and these increasing draughts, the distance between the river bed and the ships are decreasing. Rijkswaterstaat wants to know the effects of sailing at these small underkeel clearances on river beds and ship manoeuvrability. In order to quantify these effects, more knowledge about the flow field beneath sailing vessels is required, as well as the effect of the flow field on erosion of bed material. Currently no methods exist to determine the flow field beneath vessels, only a few formulations for a single maximum velocity value are available, but these are not applicable at small underkeel clearances (h / T < 1.25). Also the effect on the river bed is fairly unknown. To quantify the different effects of small underkeel clearances on the flow field physical model tests (with a length scale of 30) have been performed at Deltares. During these experiments a ship was towed through the flume, and flow velocities and pressures on the bed were measured, as well as forces on the ship. Additional experiments have been performed to investigate the effect on a moveable bed with different bed forms. From the experiments, it was found that the most important parameters that influence the flow field beneath the keel are the bow shape and the underkeel clearance. Barge bows force more flow underneath the keel than conventional bows, and this results in higher bed velocities. Decreasing keel clearances also result in significantly higher velocities at the bed. However, for very small underkeel clearances the boundary layer on the ship will interact with the boundary layer on the bed. This results in flow blockage underneath the keel. As a result, the flow needs to divert to the sides, and the velocities underneath the keel decrease. The diversion of the flow to the sides is also known as the fanning-out effect. This effect has definitely been proved by the measurements from the experiments. The effect (transverse velocities) increases with decreasing keel clearance (due to boundary layer interaction) and also increases with increasing ship widths. During the experiments, erosion of bed material was clearly observed, and its effect increased with decreasing keel clearance. However, the underkeel clearance needs to be very small (h /UKC < 1.1) to give significant bed erosion. Due to the fanning-out effect and turbulence fluctuations, most sediment transport occurred immediately alongside the vessel, rather than underneath the keel. With bed forms such as dunes the erosion increased, due to erosion at the dune tops and deposition in the troughs (10 passages of a conventional vessel over a dune resulted in a decrease in dune height of 20%). For the removal of small shoals this might be interesting, although a small underkeel clearance is necessary. Barges are preferred over conventional vessels due to the higher velocities and increased turbulence intensity. From the measured velocities during the physical model tests a model has been developed to predict the flow field underneath sailing inland navigation vessels. There are separate models for conventional vessels and for barges. The model is able to accurately predict maximum velocities (in sailing direction as well as in transverse direction), as well as a transverse velocity distribution. Compared to the previous prediction methods, the newly developed model is preferable. The results are more accurate, and the model is more extensive, due to the inclusion of transverse velocities and velocity distributions. More validation is required however, due to the lack of other data sets.Hydraulic EngineeringCivil Engineering and Geoscience
Theoretische Perspektiven in der Gesundheitssoziologie
No full textGerlinger T. Theoretische Perspektiven in der Gesundheitssoziologie. In: Kriwy P, Jungbauer-Gans M, eds. Handbuch Gesundheitssoziologie. Springer Reference Sozialwissenschaften. Wiesbaden: Springer VS; 2019: 1-21
StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis
No full textText-to-image synthesis has recently seen significant progress thanks to
large pretrained language models, large-scale training data, and the
introduction of scalable model families such as diffusion and autoregressive
models. However, the best-performing models require iterative evaluation to
generate a single sample. In contrast, generative adversarial networks (GANs)
only need a single forward pass. They are thus much faster, but they currently
remain far behind the state-of-the-art in large-scale text-to-image synthesis.
This paper aims to identify the necessary steps to regain competitiveness. Our
proposed model, StyleGAN-T, addresses the specific requirements of large-scale
text-to-image synthesis, such as large capacity, stable training on diverse
datasets, strong text alignment, and controllable variation vs. text alignment
tradeoff. StyleGAN-T significantly improves over previous GANs and outperforms
distilled diffusion models - the previous state-of-the-art in fast
text-to-image synthesis - in terms of sample quality and speed.Comment: Project page: https://sites.google.com/view/stylegan-t
Data augmentation with GANs applied to healthcare
No full textThis dissertation explores the application of Generative Adversarial Networks (GANs)
to generate time-series data, with a particular focus on Electrocardiogram (ECG) signals
used for arrhythmia detection. Data scarcity in medical fields is compounded by privacy
regulations, the technical complexities of data collection, and the rarity of certain
pathologies, all of which limit access to comprehensive datasets.
With a foundation in the MIT-BIH Arrhythmia Database, this study leverages a
Wasserstein GAN with Gradient Penalty (WGAN-GP) architecture and changes the
model’s structure by adding bidirectional Long Short-Term Memory (LSTM) layers to
generate realistic synthetic ECG signals. These synthetic signals aim to balance datasets
for arrhythmia classification, improving classifier performance where traditional Data
Augmentation (DA) methods fall short due to privacy, rarity, and complexity constraints
in medical data.
The GAN model’s training was evaluated using a combination of quantitative metrics
such as Euclidean Distance and Dynamic Time Warping (DTW), alongside visual
techniques like Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor
Embedding (t-SNE). Additionally, a classification model trained on augmented ECG
data demonstrated potential in addressing dataset imbalances and enhancing accuracy in
detecting arrhythmic events, demonstrating the GAN’s effectiveness in enhancing model
performance.
This work contributes to the broader field of healthcare data science. It highlights
the potential of GANs to overcome significant challenges by providing privacy-preserving,
diverse datasets that improve diagnostic model accuracy. Through this approach, GANs
offer a tool for medical research, facilitating the development of robust predictive models
while maintaining data integrity and confidentiality. The results underscore the potential
for GANs to impact, where enhanced data accessibility and diversity can significantly
improve patient outcomes in arrhythmia detection and beyond.Esta dissertação explora a aplicação de Redes Adversariais Generativas (GANs) para
gerar dados de séries temporais, com foco particular em sinais de eletrocardiograma (ECG)
usados para detecção de arritmias. A escassez de dados nas áreas médicas é agravada pelas
regulamentações de privacidade, pelas complexidades técnicas da recolha de dados e pela
raridade de certas patologias, que limitam o acesso a conjuntos de dados abrangentes.
Recorrendo `a base de dados de arritmia do MIT-BIH, este estudo aproveita uma
arquitetura Wasserstein GAN com Gradient Penalty (WGAN-GP) e altera a estrutura
do modelo adicionando camadas Long Short-Term Memory (LSTM) bidirecionais para
gerar sinais de ECG sintéticos realistas. Esses sinais sintéticos visam equilibrar conjuntos
de dados para classificação de arritmia, melhorando o desempenho do classificador onde
os métodos tradicionais de aumento de dados são insuficientes devido a restrições de
privacidade, raridade e complexidade em dados médicos.
O processo de treino do modelo GAN foi avaliado usando uma combinação de métricas
quantitativas, como Euclidean Distance e Dynamic Time Warping, juntamente com técnicas
visuais como PCA e t-SNE. Além disso, um modelo de classificação treinado com dados
de ECG aumentados demonstrou potencial na abordagem de desequilíbrios no conjunto
de dados e no aumento da precisão na detecção de eventos arrítmicos, demonstrando a
eficácia do GAN na melhoria do desempenho do modelo.
Este trabalho contribui para o campo da ciência de dados em saúde. Destaca o potencial
das GANs para superar desafios significativos, fornecendo conjuntos de dados diversos
que preservam a privacidade e melhoram a precisão do modelo de diagnóstico. Através
desta abordagem, os GANs oferecem uma ferramenta para a investigação médica, facilitando
o desenvolvimento de modelos preditivos robustos, mantendo ao mesmo tempo,
a integridade e a confidencialidade dos dados. Os resultados realçam o potencial de impacto
dos GANs, onde a maior acessibilidade e diversidade dos dados podem melhorar
significativamente os resultados dos pacientes na detecção de arritmia e muito mais
Data augmentation with GANs applied to healthcare
No full textThis dissertation explores the application of Generative Adversarial Networks (GANs)
to generate time-series data, with a particular focus on Electrocardiogram (ECG) signals
used for arrhythmia detection. Data scarcity in medical fields is compounded by privacy
regulations, the technical complexities of data collection, and the rarity of certain
pathologies, all of which limit access to comprehensive datasets.
With a foundation in the MIT-BIH Arrhythmia Database, this study leverages a
Wasserstein GAN with Gradient Penalty (WGAN-GP) architecture and changes the
model’s structure by adding bidirectional Long Short-Term Memory (LSTM) layers to
generate realistic synthetic ECG signals. These synthetic signals aim to balance datasets
for arrhythmia classification, improving classifier performance where traditional Data
Augmentation (DA) methods fall short due to privacy, rarity, and complexity constraints
in medical data.
The GAN model’s training was evaluated using a combination of quantitative metrics
such as Euclidean Distance and Dynamic Time Warping (DTW), alongside visual
techniques like Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor
Embedding (t-SNE). Additionally, a classification model trained on augmented ECG
data demonstrated potential in addressing dataset imbalances and enhancing accuracy in
detecting arrhythmic events, demonstrating the GAN’s effectiveness in enhancing model
performance.
This work contributes to the broader field of healthcare data science. It highlights
the potential of GANs to overcome significant challenges by providing privacy-preserving,
diverse datasets that improve diagnostic model accuracy. Through this approach, GANs
offer a tool for medical research, facilitating the development of robust predictive models
while maintaining data integrity and confidentiality. The results underscore the potential
for GANs to impact, where enhanced data accessibility and diversity can significantly
improve patient outcomes in arrhythmia detection and beyond.Esta dissertação explora a aplicação de Redes Adversariais Generativas (GANs) para
gerar dados de séries temporais, com foco particular em sinais de eletrocardiograma (ECG)
usados para detecção de arritmias. A escassez de dados nas áreas médicas é agravada pelas
regulamentações de privacidade, pelas complexidades técnicas da recolha de dados e pela
raridade de certas patologias, que limitam o acesso a conjuntos de dados abrangentes.
Recorrendo `a base de dados de arritmia do MIT-BIH, este estudo aproveita uma
arquitetura Wasserstein GAN com Gradient Penalty (WGAN-GP) e altera a estrutura
do modelo adicionando camadas Long Short-Term Memory (LSTM) bidirecionais para
gerar sinais de ECG sintéticos realistas. Esses sinais sintéticos visam equilibrar conjuntos
de dados para classificação de arritmia, melhorando o desempenho do classificador onde
os métodos tradicionais de aumento de dados são insuficientes devido a restrições de
privacidade, raridade e complexidade em dados médicos.
O processo de treino do modelo GAN foi avaliado usando uma combinação de métricas
quantitativas, como Euclidean Distance e Dynamic Time Warping, juntamente com técnicas
visuais como PCA e t-SNE. Além disso, um modelo de classificação treinado com dados
de ECG aumentados demonstrou potencial na abordagem de desequilíbrios no conjunto
de dados e no aumento da precisão na detecção de eventos arrítmicos, demonstrando a
eficácia do GAN na melhoria do desempenho do modelo.
Este trabalho contribui para o campo da ciência de dados em saúde. Destaca o potencial
das GANs para superar desafios significativos, fornecendo conjuntos de dados diversos
que preservam a privacidade e melhoram a precisão do modelo de diagnóstico. Através
desta abordagem, os GANs oferecem uma ferramenta para a investigação médica, facilitando
o desenvolvimento de modelos preditivos robustos, mantendo ao mesmo tempo,
a integridade e a confidencialidade dos dados. Os resultados realçam o potencial de impacto
dos GANs, onde a maior acessibilidade e diversidade dos dados podem melhorar
significativamente os resultados dos pacientes na detecção de arritmia e muito mais
An Assessment of GANs for Identity-related Applications
No full textInternational audienceGenerative Adversarial Networks (GANs) are now capable of producing synthetic face images of exceptionally high visual quality. In parallel to the development of GANs themselves, efforts have been made to develop metrics to objectively assess the characteristics of the synthetic images, mainly focusing on visual quality and the variety of images. Little work has been done, however, to assess overfitting of GANs and their ability to generate new identities. In this paper we apply a state of the art biometric network to various datasets of synthetic images and perform a thorough assessment of their identity-related characteristics. We conclude that GANs can indeed be used to generate new, imagined identities meaning that applications such as anonymisation of image sets and augmentation of training datasets with distractor images are viable applications. We also assess the ability of GANs to disentangle identity from other image characteristics and propose a novel GAN triplet loss that we show to improve this disentanglement
Handwritten biographical information on Paulina T. McClung Merritt
No full textA handwritten biography of Paulina T. McClung Merritt by an unknown author, 1892.
Heterogeneous and tissue-specific regulation of effector T cell responses by IFN-gamma during Plasmodium berghei ANKA infection.
No full textIFN-γ and T cells are both required for the development of experimental cerebral malaria during Plasmodium berghei ANKA infection. Surprisingly, however, the role of IFN-γ in shaping the effector CD4(+) and CD8(+) T cell response during this infection has not been examined in detail. To address this, we have compared the effector T cell responses in wild-type and IFN-γ(-/-) mice during P. berghei ANKA infection. The expansion of splenic CD4(+) and CD8(+) T cells during P. berghei ANKA infection was unaffected by the absence of IFN-γ, but the contraction phase of the T cell response was significantly attenuated. Splenic T cell activation and effector function were essentially normal in IFN-γ(-/-) mice; however, the migration to, and accumulation of, effector CD4(+) and CD8(+) T cells in the lung, liver, and brain was altered in IFN-γ(-/-) mice. Interestingly, activation and accumulation of T cells in various nonlymphoid organs was differently affected by lack of IFN-γ, suggesting that IFN-γ influences T cell effector function to varying levels in different anatomical locations. Importantly, control of splenic T cell numbers during P. berghei ANKA infection depended on active IFN-γ-dependent environmental signals--leading to T cell apoptosis--rather than upon intrinsic alterations in T cell programming. To our knowledge, this is the first study to fully investigate the role of IFN-γ in modulating T cell function during P. berghei ANKA infection and reveals that IFN-γ is required for efficient contraction of the pool of activated T cells
- …
