Find research datasets worth reusing

opencc-by-4.0Apr 2022View details →

Supplementary data release for "Cosmology and modified gravitational wave propagation from binary black hole population models"

We release the data products associated to the paper <a href="https://arxiv.org/abs/2112.05728">"Cosmology and modified gravitational wave propagation from binary black hole population models", </a><a href="https://journals.aps.org/prd/abstract/10.1103/PhysRevD.105.064030">Phys.Rev.D 105 (2022) 6 </a>. The data can be used in conjunction with the code <a href="https://github.com/CosmoStatGW/MGCosmoPop">MGCosmoPop</a> to reproduce the results of the paper.  The data product contains the following folders: * injections_GWTC3: injections used to analyze the GWTC3 catalog, generated with the code <a href="https://github.com/CosmoStatGW/MGCosmoPop">MGCosmoPop</a> . Injections are available separately for O1-O2, O3a, O3b for minimum SNR of 10, 11, 12 (folder names are self-explicative). Each folder contains a file named selected.h5 with the injections. For loading them, refer to the tutorial of the code <a href="https://github.com/CosmoStatGW/MGCosmoPop">MGCosmoPop</a> . * mock_BPL_5yr_GR : mock data for 5 years of aLIGO observations, with fiducial cosmological model set to General Relativity (see the paper for details) * mock_BPL_5yr_MG : mock data for 5 years of aLIGO observations, with fiducial cosmological model set to a modified gravity model with modified gravitational-wave propagation (see the paper for details) * injections_mock : injections for analyzing the mock datasets above

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - CIFAR10

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on CIFAR10. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "cifar_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained with small and large CNN models, in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - STL10 - Raw Datasets

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from STL10. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains the raw model zoos as collections of models (file names beginning with "cifar_"). Zoos are trained with small and large CNN models, in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). Due to the large filesize, the preprocessed datasets are hosted in a separate repository. The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - SVHN

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from SVHN. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "svhn_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - Fashion-MNIST

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from Fashion-MNIST. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "fmnist_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

Model Zoo: A Dataset of Diverse Populations of Neural Network Models - MNIST

Abstract In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before. Dataset This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from MNIST. All zoos with extensive information and code can be found at www.modelzoos.cc. This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "mnist_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models. For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

opencc-by-4.0Jan 2018View details →

Supplementary material for "Palaeo-demographic modelling supports a population bottleneck during the Pleistocene-Holocene transition in Iberia"

This Supplementary material encompasses the archaeological radiocarbon dates for the Iberian Peninsula analyzed in the article “Palaeo-demographic modelling supports a population bottleneck during the Pleistocene-Holocene transition in Iberia". Specifically: <ul> <li>‘PALEODEM_dataset.csv’ provides the dataset in comma-separated format. Contains the compiled raw data of each radiocarbon date used by every analytical procedure.</li> <li>‘PALEODEM_sites.xlsx’ is a spreadsheet containing the basic information of each archaeological site analyzed in this work. The columns include ID_site, name of Archaeological site, no of 14C dates, no of assemblages, Regional Unit of correspondence, and bibliographical Reference.</li> <li>‘REFERENCES for PALEODEM_sites.docx’ is a text archive that collects the bibliographical references listed in the above mentioned  ‘Reference’ column, that have been the main sources of archaeological information.</li> </ul>

opencc-by-4.0Oct 2019View details →

Combining statistical and mechanistic models to unravel the drivers of mortality within a rear-edge beech population - Supporting Material

Supporting material for the study: "Combining statistical and mechanistic models to unravel the drivers of mortality within a rear-edge beech population." Authors: Cathleen Petit-Cailleux1, Hendrik Davi1, François Lefèvre1, Joseph Garrigue2, Jean-André Magdalou2, Christophe Hurson2,3, Elodie Magnanou2,4, and Sylvie Oddou-Muratorio1.   Adresses 1INRA, UR 629 Ecologie des Forêts Méditerranéennes, URFM, Avignon, France 2Réserve Naturelle Nationale de la Forêt de la Massane, France 3Fédération des Réserves Naturelles Catalanes, Prades, France 4Sorbonne Université, CNRS, Biologie Intégrative des Organismes Marins, BIOM, F-66650 Banyuls-sur-Mer, France ORCID: Cathleen Petit-Cailleux: <a href="https://orcid.org/0000-0001-7714-6583">https://orcid.org/0000-0001-7714-6583</a> François Lefèvre : <a href="https://orcid.org/0000-0003-2242-7251">https://orcid.org/0000-0003-2242-7251</a> Sylvie Oddou-Muratorio <a href="https://orcid.org/0000-0003-2374-8313">https://orcid.org/0000-0003-2374-8313</a>   ------------- Raw data of the Table_Massane_moratlity_trees.csv and climate can be obtained from Joseph Garrigue, Jean-André Magdalou and Christophe Hurson. The inventories files and daily climate are the input dataset to run CASTANEA models. All details are provided in the article.

opencc-by-4.0May 2023View details →

Population-level health and economic impacts of introducing Vaccae vaccination in China: A modeling study

Supplementary to "Population-level health and economic impacts of introducing Vaccae vaccination in China: A modeling study"

opencc-by-4.0Jan 2023View details →

Open-population models for estimating roadkill rates - Data and R Code

Roadkill carcass capture-recapture data, capture histories for four and eight-occasion designs, and R code (with JAGS code) for roadkill rates estimation.

zenodo40/100

Magellan/M2FS Spectroscopy of Galaxy Clusters: Stellar Population Model and Application to Abell 267

supplementary data products, including all sky-subtracted spectra from individual galaxies, as well as random draws from posterior PDFs for model parameters (see included README file)

opencc-by-4.0Jul 2017View details →

opencc-zeroJan 2024View details →

An integrated population model reveals source-sink dynamics for competitively subordinate African wild dogs linked to anthropogenic prey depletion

<ol> <li>Many African large carnivore populations are declining due to decline of the herbivore populations on which they depend. The densities of apex carnivores like the lion and spotted hyena correlate strongly with prey density, but competitive subordinates like the African wild dog benefit from competitive release when the density of apex carnivores is low, so the expected effect of a simultaneous decrease in resources and dominant competitors is not obvious. </li> <li>Wild dogs in Zambia's Luangwa Valley Ecosystem occupy four ecologically similar areas with well-described differences in the densities of prey and dominant competitors, due to spatial variation in illegal offtake.</li> <li>We used long-term data to fit a Bayesian integrated population model (IPM) of the demography and dynamics of wild dogs in these four regions. The IPM used Leslie projection to link a Cormack-Jolly-Seber model of area-specific survival (allowing for individual heterogeneity in detection), a zero-inflated Poisson model of area-specific fecundity, and a state-space model of population size that used estimates from a closed mark-capture model as the counts from which (latent) population size was estimated.</li> <li>The IPM showed that both survival and reproduction were lowest in the region with the lowest density of preferred prey (puku, Kobus vardonii, and impala, Aepyceros melampus), despite little use of this area by lions. Survival and reproduction were highest in the region with the highest prey density, and intermediate in the two regions with intermediate prey density. The population growth rate (λ) was positive for the population as a whole, strongly positive in the region with the highest prey density, and strongly negative in the region with the lowest prey density.</li> <li>It has long been thought that the benefits of competitive release protect African wild dogs from the costs of low prey density. Our results show that the costs of prey depletion overwhelm the benefits of competitive release and cause local population decline where anthropogenic prey depletion is strong. Because competition is important in many guilds and humans are affecting resources of many types, it is likely that similarly fundamental shifts in population limitation are arising in many systems.</li> </ol>

opencc-zeroMay 2022View details →

Integrated population model for the Mallard in the Netherlands

Europe's highest densities of breeding Mallards (Anas platyrhynchos) are found in the Netherlands, but the breeding population there has declined by ~30% since the 1990s. The exact cause of this decline has remained unclear. Here, we used an integrated population model to jointly analyze Mallard population survey, nest survey, duckling survival and band-recovery data. We used this approach to holistically estimate all relevant vital rates, including duckling survival rates for years for which no explicit data were available. Mean vital rate estimates were high for nest success (0.38 ±0.01) and egg hatch rate (0.96 ±0.001), but relatively low for clutch size (8.2 ±0.05) compared to populations in other regions. Estimates for duckling survival rate for the three years for which explicit data were available were low (0.16-0.27) compared to historical observations, but were comparable to rates reported for other regions with declining populations. Finally, mean survival rate was low for ducklings (0.18 ±0.02), but high and stable for adults (0.71 ±0.03). Population growth rate was only affected by variation in duckling survival, but since this is a predominantly latent state variable, this result should be interpreted with caution. However, it does strongly indicate that none of the other vital rates, all of which were supported by data, was able to sufficiently explain the population decline. Together with a comparison with historic vital rates, these findings point to a reduced duckling survival rate as the likely cause of the decline. Candidate drivers of reduced duckling survival are increased predation pressure and reduced food availability, but this requires future study. Integrated population modeling can provide valuable insights into population dynamics even when empirical data for a key parameter are partly missing.

opencc-zeroMay 2022View details →

Gene drives for vertebrate pest control: realistic spatial modelling of eradication probabilities and times for island mouse populations

Invasive alien species continue to threaten global biodiversity. CRISPR-based gene drives, which can theoretically spread through populations despite imparting a fitness cost, could be used to suppress or eradicate pest populations. We develop an individual-based, spatially explicit, stochastic model to simulate the ability of CRISPR-based homing and X-chromosome shredding drives to eradicate populations of invasive mice (Mus muculus) from islands. Using the model, we explore the interactive effect of the efficiency of the drive constructs and the spatial ecology of the target population on the outcome of a gene-drive release. We also consider the impact of polyandrous mating and sperm competition, which could compromise the efficacy of some gene-drive strategies. Our results show that both drive strategies could be used to eradicate large populations of mice. Whereas parameters related to drive efficiency and demography strongly influence drive performance, we find that sperm competition following polyandrous mating is unlikely to impact the outcome of an eradication effort substantially. Assumptions regarding the spatial ecology of mice influenced the probability of and time required for eradication, with short-range dispersal capabilities and limited mate-search areas producing `chase' dynamics across the island characterised by cycles of local extinction and recolonization by mice. We also show that highly efficient drives are not always optimal, when dispersal capabilities are low, rapid local population supression around the introduction sites can cause loss of the gene drive before it can spread to the entire island. We conclude that, although the design of efficient gene drives is undoubtedly critical, accurate data on the spatial ecology of target species is critical for predicting the result of a gene-drive release.

zenodo40/100

Data associated with the manuscript "Simple statistical models can be sufficient for testing hypotheses with population time series data"

This is a revised version of the archive of R code and data used in the manuscript, Simple statistical models can be sufficient for testing hypotheses with population time series data. The data are in three files. etodata1.csv and etodata2.csv contain two versions of the same data for shoal-dwelling fishes in the Etowah River and associated environmental covariates. knz_dat contains data for small mammals collected in the Konza Prairie Biological Station and associated environmental covariates. The R code consists of four primary files that call nine auxiliary files. CaseStudy1-main_code and CaseStudy2-main_code are the primary files for running the two case studies. Simulations1 and Simulations2 are the files for running the two batteries of simulations. We thank the Konza Prairie Biological Station and Konza Prairie Long-Term Ecological Research Program supported by the National Science Foundation (DEB-1440484) for collecting and providing access to mammal community data. More details are in the manuscript and supporting information.

opencc-by-4.0Sep 2021View details →

opencc-zeroJun 2022View details →

Code: A model of wild bee populations accounting for spatial heterogeneity and climate induced temporal variability of food resources at the landscape level

The viability of wild bee populations and the pollination services that they provide are driven by the availability of food resources during their activity period and within the surroundings of their nesting sites. Changes in climate and land use influence the availability of these resources and are major threats to declining bee populations. Because wild bees may be vulnerable to interactions between these threats, spatially explicit models of population dynamics that capture how bee populations jointly respond to land use at a landscape scale and weather are needed. Here, we developed a spatially and temporally explicit theoretical model of wild bee populations aiming for a middle ground between the existing mapping of visitation rates using foraging equations and more refined agent-based modelling. The model is developed for Bombus sp. and captures within-season colony dynamics. The model describes mechanistically foraging at the colony level and temporal population dynamics for an average colony at the landscape level. Stages in population dynamics are temperature-dependent triggered with a theoretical generalized seasonal progression, which can be informed by growing degree days (GDD). The purpose of the LandscapePhenoBee model is to evaluate the impact of systematic changes and within-season variability in resources on bee population sizes and crop visitation rates. In a simulation study, we used the model to evaluate the impact of the shortage of food resources in the landscape arising from extreme drought events in different types of landscapes (ranging from different proportions of semi-natural habitats and early and late flowering crops) on bumblebee populations.