Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
209
datasets available to search
ShareScore release 0.9.0
Dataset results
209 results for “probabilistic”
Probabilistic Approach for the Fatigue Strength Prediction of Polymers
<p>The dominating factor in the fatigue of structures made from fiber reinforced polymers (FRP), for example wind turbine blades, is the polymer matrix. Traditionally, experimental stress-life data of polymers is approximated via a linear double-log Basquin model. Recently, the non-linear stress-life formulation by Stüssi was found to provide a better fit of the experimental data with a substantially reduced standard deviation. Moreover, a non-linear constant-life formulation, as proposed by Boerstra, for example, can enhance the representation of the mean stress effect compared to state-of-the-art linear models, i.e., the modified Goodman relation. To this end, we incorporated Stüssi’s model into the Boerstra relation to take account of the mean stress effects of an epoxy. This stress-life formulation was then enhanced with the Weibull probability function. The probabilistic-stress-life model provided a good approximation of the fatigue performance as a function of the stress ratio on the basis of an experimental data set. Finally, we suggested a step-wise engineering approach to derive the permissible stress-life with a view to practical design purposes. The procedure increased the reliability of the fatigue design evaluation compared to the state-of-the-art methodologies.</p> <p>Slides: <a href="https://doi.org/10.5281/zenodo.4443149">https://doi.org/10.5281/zenodo.4443149</a><br> Paper: <a href="https://doi.org/10.2514/6.2021-1289">https://doi.org/10.2514/6.2021-1289</a></p>
Data from: Stimulus discriminability may bias value-based probabilistic learning
Reinforcement learning tasks are often used to assess participants' tendency to learn more from the positive or more from the negative consequences of one's action. However, this assessment often requires comparison in learning performance across different task conditions, which may differ in the relative salience or discriminability of the stimuli associated with more and less rewarding outcomes, respectively. To address this issue, in a first set of studies, participants were subjected to two versions of a common probabilistic learning task. The two versions differed with respect to the stimulus (Hiragana) characters associated with reward probability. The assignment of character to reward probability was fixed within version but reversed between versions. We found that performance was highly influenced by task version, which could be explained by the relative perceptual discriminability of characters assigned to high or low reward probabilities, as assessed by a separate discrimination experiment. Participants were more reliable in selecting rewarding characters that were more discriminable, leading to differences in learning curves and their sensitivity to reward probability. This difference in experienced reinforcement history was accompanied by performance biases in a test phase assessing ability to learn from positive vs. negative outcomes. In a subsequent large-scale web-based experiment, this impact of task version on learning and test measures was replicated and extended. Collectively, these findings imply a key role for perceptual factors in guiding reward learning and underscore the need to control stimulus discriminability when making inferences about individual differences in reinforcement learning.
Evaluating the performance of probabilistic algorithms for phylogenetic analysis of big morphological datasets: a simulation study
<p>Reconstructing the tree of life is an essential task in evolutionary biology. It demands accurate phylogenetic inference for both extant and extinct organisms, the latter being almost entirely dependent on morphological data. While parsimony methods have traditionally dominated the field of morphological phylogenetics, a rapidly growing number of studies are now employing probabilistic methods (maximum likelihood and Bayesian inference). The present-day toolkit of probabilistic methods offers varied software with distinct algorithms and assumptions for reaching global optimality. However, benchmark performance assessments of different software packages for the analyses of morphological data, particularly in the era of big data, are still lacking. Here, we test the performance of four major probabilistic software under variable taxonomic sampling and missing data conditions: the Bayesian inference-based programs MrBayes and RevBayes, and the maximum likelihood-based IQ-TREE and RAxML. We evaluated software performance by calculating the distance between inferred and true trees using a variety of metrics, including Robinson-Foulds (RF), Matching Splits (MS), and Kuhner-Felsenstein (KF) distances. Our results show that increased taxonomic sampling improves accuracy, precision, and resolution of reconstructed topologies across all tested probabilistic software applications and all levels of missing data. Under the RF metric, Bayesian inference applications were the most consistent, accurate, and robust to variation in taxonomic sampling in all tested conditions, especially at high levels of missing data, with little difference in performance between the two tested programs. The MS metric favored more resolved topologies that were generally produced by IQ-TREE. Adding more taxa dramatically reduced performance disparities between programs. Importantly, our results suggest that the RF metric penalizes incorrectly resolved nodes (false positives) more severely than the MS metric, which instead tends to penalize polytomies. If false positives are to be avoided in systematics, Bayesian inference should be preferred over maximum likelihood for the analysis of morphological data.</p>
Data from: Probabilistic interfractional motion carbon ion radiation therapy dose distribution for prostate cancer shows rectum sparing with moderate target coverage degradation
Purpose: This observational study investigates the influence of interfractional motion on clinical target volume (CTV) coverage, planning target volume (PTV) margins, and rectum tissue sparing in carbon ion radiation therapy (CIRT). It reports dose coverage to target structures and organs at risk in the presence of interfractional motion, investigates rectal tissue sparing, and provides recommendations for further lowering the rate of toxicity. We also propose probabilistic DVH for consideration in treatment planning to represent probable dose to the clinic's patient population. Methods: At Gunma University Hospital intensity-modulated x-ray therapy (IMXT, aka IMRT) prostate cancer patients are positioned on a table which is shifted twice based on cone-beam computed tomography (CBCT) to align bones and then align prostate tissue to isocenter. These shifts thereby contain interfractional motion. 1306 such tableshifts from 85 patients were collected. Normal probability distributions were fit to the difference between bone-matching and prostate-matching CBCT-to-planning CT tableshifts (i.e. interfractional motion). Between 2011 and 2016 CIRT prostate patients were treated with PTV1 and PTV2 margins as follows: PTV1 extends the prostate contour by 10/10, 5/10, 6/6 mm in the right/left, posterior/anterior, and superior/inferior directions, respectively, and the proximal seminal vesicles contour by 5 mm superiorly and inferiorly, 3 mm right and left. PTV2 reduces PTV1 posteriorly along a straight line to exclude the rectum and reduces the superior and inferior margins by 6 mm. From those treated with these margins, 40 patients' beam data were selected to create probable interfractional motion: The previously fit normal probability distributions were randomly sampled 2000 times per patient and beams shifted to simulate this motion. These shifted dose distributions were scaled down proportionately in magnitude and summed to obtain probable blurred dose distributions. Results: Probable dose to rectum is substantially less than planned for doses higher than 10 Gy(RBE). Absolute DVH show that mean clinical target volumes are about 138-670 cm3 smaller for a given probable dose than planned doses higher than 57 Gy(RBE) after accounting for standard error. Cumulative DVH show mean CTV fraction receiving a given probable dose is less than the mean fraction receiving the corresponding planned dose for doses larger than 52 Gy(RBE), up to 19% less at 57.4 Gy(RBE). Our PTV1 margins generally cover 95% of interfractional motion but seminal vesicles and inferior prostate receive less dose than planned due to insufficient PTV2 margins. Conclusion: Assuming rigidly shifting interfractional motion around the prostate region and neglecting minor changes in soft tissue stopping power, interfractional motion resulted in underdosing or tissue sparing in all cases. Given our low rates of relapse and recurrence, it appears less curative dose is needed than previously thought or else the target may be smaller than previously thought. In-room CT may be useful to lower dose, shrink target margins, conduct PET auto-activation dose verification studies and account for interfractional motion.
Dataset for A probabilistic forecast methodology for volatile electricity prices in the Australian National Electricity Market
Open the record for dataset details and reuse information.
North Atlantic Ocean Drifter Dataset for Multivariate Probabilistic Regression with Natural Gradient Boosting
<p>Dataset created for Multivariate Probabilistic Regression with Natural Gradient Boosting application section. Supplied for replication. Variables in the dataset:</p> <ul> <li>Tx, Ty, Wx, Wy: Wind Stress (Pa) (T), Wind Speed (m/s) (W) in the longitudinal (x) and latitudinal (y) directions.</li> <li>u_av, v_av: surface geostrophic sea water velocity (m/s) longitudinal and latitudinal respectively.</li> <li>lon,lat: The longitude (-180,180) and latitude (-90,90) of the drifters.</li> <li>t: day of year (0,366).</li> <li>u, v: The velocities of the drifters (m/s).</li> </ul> <p>Note u, v, Tx, Ty, Wx, Wy are all filtered as described in the papers supplemental information.</p> <p>If used the following must be acknowledged or cited appropriately. You must follows the terms of the use of the following sources:</p> <p>The Global Drifter Program</p> <p>https://www.aoml.noaa.gov/phod/gdp/interpolated/data/all.php</p> <p>(lon, lat, u, v)</p> <p>CMEMS datasets which are interpolated to the drifter locations:</p> <p>1. GLOBAL OCEAN GRIDDED L4 SEA SURFACE HEIGHTS AND DERIVED VARIABLES REPROCESSED (COPERNICUS CLIMATE SERVICE) https://resources.marine.copernicus.eu/product-detail/SEALEVEL_GLO_PHY_CLIMATE_L4_REP_OBSERVATIONS_008_057/INFORMATION<br> (u_av, v_av)</p> <p>2. Global Ocean Wind L4 Reprocessed 6 hourly Observations https://resources.marine.copernicus.eu/product-detail/WIND_GLO_WIND_L4_REP_OBSERVATIONS_012_006/INFORMATION (Wx, Wy, Tx, Ty)<br> <br> <br> <strong>Note</strong> this dataset will not be regularly updated. Any changes in the source products will not be brought into this dataset.</p>
SixthSense: Debugging Convergence Problems in Probabilistic Programs via Program Representation Learning
<p>This is a dataset for our paper: "SixthSense: Debugging Convergence Problems in Probabilistic Programs via Program Representation Learning" published at FASE 2022. Find more details at https://github.com/uiuc-arc/sixthsense</p>
Supplementary material of "VIGIL: A Python tool for automatized probabilistic VolcanIc Gas dIspersion modeLling"
<p>This dataset includes the Supplementary material of the paper: </p> <p>Dioguardi F, Massaro S, Chiodini G, Costa A, Folch A, Macedonio G, Sandri L, Selva J, Tamburello G. VIGIL: A Python tool for automatized probabilistic VolcanIc Gas dIspersion modeLling. Ann. Geophys. [Internet]. 2022Apr.3 [cited 2022Apr.7];65(1). Available from: https://www.annalsofgeophysics.eu/index.php/annals/article/view/8796</p>
A Probabilistic Framework for Mutation Testing in Deep Neural Networks - Models archive Part 2
<p>Models used as part of the paper "A Probabilistic Framework for Mutation Testing in Deep Neural<br> Networks ?" submitted to the journal Information and Software Technology</p> <p>Replication package using the data is available at https://github.com/FlowSs/PM</p>
A Probabilistic Framework for Mutation Testing in Deep Neural Networks - Models archive Part 3
<p>Models used as part of the paper "A Probabilistic Framework for Mutation Testing in Deep Neural<br> Networks ?" submitted to the journal Information and Software Technology</p> <p>Replication package using the data is available at https://github.com/FlowSs/PMT</p>
A Probabilistic Framework for Mutation Testing in Deep Neural Networks - Models archive Part 1
<p>Models used as part of the paper "A Probabilistic Framework for Mutation Testing in Deep Neural<br> Networks ?" submitted to the journal Information and Software Technology</p> <p>Replication package using the data is available at https://github.com/FlowSs/PMT</p>
Training data - Hydraulic Head Probabilistic MLP-NN
<p>Training data for paper on hydraulic head predictions using an MLP-NN.</p>
In vivo probabilistic atlas of white matter tracts of the human subthalamic area combining track density imaging and optimized diffusion tractography
<p>Sample code to reproduce the results of the paper "In vivo probabilistic atlas of white matter tracts of the human subthalamic area combining track density imaging and optimized diffusion tractography", by Basile et al., Brain Structure and Function 2022. </p> <p>The repository also includes the probabilistic atlas of the human subthalamic area.</p>
Supporting data for: Planted conifer seedling survival in a post-wildfire landscape in New Mexico: From experimental planting to predictive probabilistic surfaces
<p>Across the southwestern United States, high-severity wildfire is causing increasingly large patches of tree mortality, removing the seed sources required for the natural regeneration of these formerly conifer-dominated landscapes. Planting tree seedlings can accelerate succession and restore the ecosystem services that pre-wildfire forests provided, but in the semi-arid southwestern US, the survival of planted conifer seedlings is typically low. We present data on a seedling planting experiment within the footprint of the 2011 Las Conchas Fire in northern New Mexico, using microclimate variables and site-level topographic indices to determine which factors influence seedling survival of four conifer tree species. We then extrapolate these findings to the wider landscape for the most commonly planted species, producing a spatial projection of probabilistic seedling survival as a function of both site- and landscape-scale factors, using readily available topographic data. Both sets of analyses highlight the importance of landscape heterogeneity, quantifiable using topographic indices, in shaping microclimatic environments and affecting seedling survival in a predictable manner. Our results demonstrate that using these methods to guide future plantings may increase conifer tree seedling survival in high-severity post-wildfire landscapes.</p>
Dataset and results for "Comparing machine learning and deep learning models for probabilistic post-processing of satellite precipitation-driven streamflow simulation"
<p>Dataset and results for "Comparing machine learning and deep learning models for probabilistic post-processing of satellite precipitation-driven streamflow simulation"</p> <p>Yuhang Zhang1, Aizhong Ye1*, Phu Nguyen2, Bita Analui2, Soroosh Sorooshian2, Kuolin Hsu2</p> <p>1 State Key Laboratory of Earth Surface Processes and Resource Ecology, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China.</p> <p>2 Center for Hydrometeorology and Remote Sensing, Department of Civil and Environmental Engineering, University of California, Irvine, Irvine, California, CA 92697, USA.</p> <p>## Dataset </p> <p>Streamflow simulations from one observed precipitation (CMA) and three satellite precipitation products (PDIR, IMERG-F, and GSMaP) for 522 sub-basins.</p> <p>- Q-CMA (streamflow reference)<br> - Q-PDIR (uncorrected)<br> - Q-IMERGF (uncorrected)<br> - Q-GSMAP (uncorrected)</p> <p>### Data structure</p> <p>- Head section (row1-row5)<br> - SubNO: 522 <br> - BeginT: 2003-01-01 00:00 <br> - EndT: 2019-12-31 00:00 <br> - Interval: 1440s (daily)<br> - Revise: 10 (scaling factor to keep int datatype)<br> - Point1 Point2 ... (Subbasin No.)<br> - Data section<br> - 6209 rows, 522 cols</p> <p>## Results</p> <p>Two post-processing model results for test period (2015-1-1 to 2018-12-31).</p> <p>### Data structure</p> <p>- 1462 rows, every row denotes each day from 2015-1-1 to 2018-12-31</p> <p>- 100 columns, every column denotes each quantile from 0.005 to 0.995, total 100 quantiles.</p> <p>### qrf-output</p> <p>- pdir (single input)<br> - imergf (single input)<br> - gsmap (single input)<br> - all (multiple inputs)</p> <p>### lstm-output</p> <p>- pdir (single input)<br> - imergf (single input)<br> - gsmap (single input)<br> - all (multiple inputs)</p> <p> </p>
Data for "Diffusion Tempering Improves Parameter Estimation with Probabilistic Integrators for Ordinary Differential Equations"
Open the record for dataset details and reuse information.
Dataset: Probabilistic work extraction on a classical oscillator beyond the second law
<p>Data set for the article "Probabilistic work extraction on a classical oscillator beyond the second law" , currently published as a preprint on arXiv, and accepted in Physical Review Letters.</p> <p>The preprint and SM have been uploaded for convenience.</p> <p>Inside the compressed file, you will find several folders</p> <p>It contains:</p> <p> </p> <ol> <li><strong>/Exemple trajectories/</strong> <ul> <li> 10 folders named Protocol 0 , Protocol 1 .... each containing the raw data of the 100 first experiments for each protocol. Each protocol was repeated 3000 time to increase the statistics for the article. We provide a representative subset.</li> <li>The Python analysis code "readexperimentalfile.py" . It reads the data in the aforementioned files, and computes the initial potential, the final potential, and the work associated to each protocols. Optionnaly, it saves trajectories used create Fig3 and Fig 4.</li> </ul> </li> <li><strong>/Figures/</strong> <ul> <li>Fig 1 : The pdf file for Fig1. (stochastic2.pdf) as presented in the article. Other formats can be provided on demand.</li> <li>Fig 2: The pdf file for Fig2, Setup.pdf</li> <li>Fig 3: The python code used to create Fig3 , named "plotpotentiel.py". To use it, one need the parameters tuning the initial potential used in the fifth protocol 'fitdepart5.npy', the final potential 'fitfin5.npy' as well as the set of trajectories "zfin5.npy" and "zinit5.npy". All those files were created using the provided python code "readexperimentalfile.py" , and are already provided in the same folder.</li> <li>Fig4: The python code used to create Fig4, named "plottrajex.py". "ztotsubset5.npy" and "lambdasubset5.npy" are provided to directly plot z(t) and lambda(t) during the protocol. "fitdepart5.npy" and 'fitfin5.npy", also created using "readexperimentalfile.py", are needed. </li> <li>Fig5: a) Everything needed to plot the first part of Fig 5. using the python file "workdpdf.py". All the datas of the other experiments are provided, and one can also choose to plot the work probability density function for other protocols (by changing the 5 t o another number. b) "mainfig.py" creates the bottom part of Fig5.</li> </ul> </li> </ol>
Datasets for "Scalable interpolation of satellite altimetry data with probabilistic machine learning"
<p>Elevation (radar freeboard and sea-level anomaly) fields from CryoSat-2, Sentinel-3A, and Sentinel-3B, over the period December 1st 2018 - April 30th 2019. These data were processed for the Arctic domain using the European Space Agency's Grid Processing on Demand (GPOD) service. Processing follows the steps outlined in Lawrence et al., 2021 (<a title="Persistent link using digital object identifier" href="https://doi.org/10.1016/j.asr.2019.10.011" target="_blank" rel="noreferrer noopener">https://doi.org/10.1016/j.asr.2019.10.011</a>). These data are provided at along-track, 5 km and 50 km resolution, where gridded data follow the EASE grid definition (<a href="https://doi.org/10.3390/ijgi1010032">https://doi.org/10.3390/ijgi1010032</a>).</p> <p>These data were used to develop the open-source Python programming library GPSat (https://github.com/CPOMUCL/GPSat), which uses local Gaussian Process models to perform scalable interpolation of non-stationary satellite altimetry data. The 'Source_data.xlsx' file contains the data corresponding to figures in the published Nature Communications article 'Scalable interpolation of satellite altimetry data with probabilistic machine learning'.</p>
Probabilistic Volcanic Ash Uncertainty
<h1>Probabilistic Volcanic Ash Uncertainty</h1> <p>Incorporating source parameter (MER and plume height) alongside meteorological variability in volcanic ash hazard dispersion forecasting. Contains code and supporting data submitted alongside manuscript "Incorporating Source Parameter and Meteorological Variability in the Generation of Probabilistic Volcanic Ash Hazard Forecasts" to Journal of Geophysical Research: Atmospheres. Requires The Met Office's <a href="https://www.metoffice.gov.uk/research/approach/modelling-systems/dispersion-model">Numerical Atmospheric-dispersion Modelling Environment</a> (NAME), which is available by licence from the UK Met Office.</p> <h2>Introduction</h2> <p>Airborne volcanic ash is hazardous for aircraft. To manage this risk, Volcanic Ash Advisory Centres (VAACs) provide forecasts of ash clouds following a volcanic eruption. These forecasts are created using dispersion models that predict the transport of ash based on eruption details and weather data. These sets of inputs have large uncertainties that can affect the accuracy of the forecasts. The paper this project is associated with presents a method for producing probabilistic forecasts that account for these uncertainties. Typically, weather uncertainties are handled by using multiple weather predictions, referred to as an ensemble. Dispersion outcomes depend on the eruption plume height and mass eruption rate (MER), which are related but have large associated uncertainties. Our method uses a statistical approach to incorporate these uncertainties into forecasts to allow for the calculation of probabilities of different ash concentration levels for aviation. It uses the same number of model runs as there are ensemble members, and does not require eruption details (plume height, MER, and emission profile) to be specified in advance, making it a computationally efficient approach as the bulk of computations can be done after a small number of initial model runs.</p> <h2>Contents</h2> <p>The zip file consists of four folders: analysis, notebooks, pvauncertainty, and scripts.</p> <h3>analysis</h3> <p>Contains two sub-folders:</p> <ul> <li>fig-scripts: scripts for post-processing of data and generation of figures for the submitted manuscript.</li> <li>data: post-processed output data of NAME simulations.</li> </ul> <h3>notebooks</h3> <p>The Jupyter notebooks get-started-pt1 and get-started-pt2 illustrate how the package can be used with NAME. Users must provide their own NAME input files to simulate volcanic ash dispersion; minimal non-working examples of code block segments that must be changed are given in scripts.</p> <h3>pvauncertainty</h3> <p>The Python package contains classes to set up volcanic ash simulations in NAME and evaluate resultant probabilistic quantities:</p> <ul> <li>Set up NAME inputs for a volcanic ash emission given a plume height observation, or range for the height: <ul> <li>Using deterministic or ensemble meteorology</li> <li>Provides a unit MER for later rescaling</li> <li>Given a plume height range and interval step size, initialises NAME with multiple interval emissions to be saved separately</li> <li>Sets NAME running on a SLURM environment</li> </ul> </li> <li>Evaluate probabilistic quantities of volcanic ash concentrations: <ul> <li>Conditional exceedance probabilities given ensemble member</li> <li>Conditional exceedance probabilities given plume height observation</li> <li>Percentiles of ash concentration given plume height observation</li> <li>Overall exceedance probabilities given plume height distribution (Gaussian distribution by default) by numerical integration (quadrature)</li> <li>Plotting of probabilistic quantities of volcanic ash concentrations</li> </ul> </li> </ul> <p>To install the package, navigate to its parent folder and execute "pip install -e .".</p> <h3>scripts</h3> <p>Contains scripts for setting up ensemble or deterministic NAME runs, given a csv file of plume height and MER values, and minimal example NAME input files.</p> <h2>License</h2> <p>This project is licensed under the BSD 3-Clause License.</p>
Probabilistic Model Checking for Temporal Logics in Weighted Structures - Experiments
<p>Supplementary material for the thesis</p> <p>Probabilistic Model Checking for Temporal Logics in Weighted Structures</p> <ul> <li>data set for the experiments</li> <li>results of the experiments</li> <li>evaluation scripts</li> </ul>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.