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.
921
datasets available to search
ShareScore release 0.7.1
Dataset results
921 results for “Neural Network”
PMD hypomethylation human (hg19) neural network scores
<p>Global loss of DNA methylation in mammalian genomes occurs cumulatively as a mitotic process during aging and cancer, primarily in Partially Methylated Domains (PMDs). It has been shown that local sequence context (100bp) has a strong effect on the rate of demethylation of individual CpG dinucleotides within PMDs. Here, we train a deep learning model to characterize this sequence dependence further, finding that methylation loss can be predicted from a CpG’s 150bp sequence context alone with an AUC of 0.95. We use re-methylation rates of newly synthesized DNA to show that CpGs with fast-loss sequence context are inefficiently re-methylated. Interestingly, we find that the 10% of CpGs predicted to have the “slowest” rate of loss lose almost no DNA methylation in healthy cell types. These same slow-loss CpGs lose a significant amount of DNA methylation in cancer, suggesting that they could be responsible for deregulation of genes and transposable elements that are associated with DNA hypomethylation in cancer.</p> <p>This directory contains the Nov. 18, 2020 version of the human (hg19) CpG hypomethylation Neural network scores in a single tab-delimited (bedgraph) file:<br> <strong>multitissue-nn-scores.allCGs.0based.hg19.bedgraph.gz</strong><br> with the following columns:<br> 1: chromosome (hg19)<br> 2: start coord (hg19, 0-based)<br> 3: end coord (hg19, 0-based)<br> 4: multi-tissue NN score (0-1). Close to 0 is classified as slow-loss CpG, close to 1 is classified as fast loss CpG5: Num CpGs in 150 bp window (including central CpG, so minimum is 1).</p> <p> </p> <p>The full version of the NN scores with additional details are in the file <strong>zhou-bian.allCGs.1based.hg19.tsv.gz</strong></p> <p>Each row is a CG which provides (1) chromosome, (2) the corresponding C coordinate on the forward (watson) strand of the reference genome in one-based coordinates, (3) Neural network score, (4) number of CpGs within the 150bp sequence centered on this CpG, including the center CpG, (5) CpG is within a CpG island (0, no; 1, yes), CpG is within ENCODE blacklist (0, no; 1, yes)</p> <p> Here the CpG islands are the union set of Irizarry (Irizarry et al. 2009, Nat Genet), Takai-Jones (Takai et al. 2002, PNAS), Gardner-Gardin CGIs (Gardner-Gardin et al. 1987, J Mol Biol.). The blacklist was downloaded from https://github.com/Boyle-Lab/Blacklist/tree/master/lists.</p> <p>Additional files are included here:<br> <strong>zhou_pmds.0based.hg19.bed.gz</strong>: Input PMD CpGs from the Zhou (multi-tissue) dataset<br> <strong>bian_pmds.crc01.0based.hg19.bed.gz</strong>: Input PMD CpGs from the Bian (intra-tumor) dataset<br> <strong>zhou_bian_train_test_data.tar.gz</strong>: All training and test CpGs, including labels and sequence windows.</p> <p> </p> <p> </p>
Model Zoo: A Dataset of Diverse Populations of Neural Network Models - CIFAR10
<p><strong>Abstract</strong></p> <p>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.</p> <p><strong>Dataset</strong></p> <p>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.</p> <p>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.</p> <p>For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.</p>
Model Zoo: A Dataset of Diverse Populations of Neural Network Models - STL10 - Raw Datasets
<p><strong>Abstract</strong></p> <p>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.</p> <p><strong>Dataset</strong></p> <p>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.</p> <p>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.</p> <p>For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.</p>
Model Zoo: A Dataset of Diverse Populations of Neural Network Models - SVHN
<p><strong>Abstract</strong></p> <p>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.</p> <p><strong>Dataset</strong></p> <p>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.</p> <p>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.</p> <p>For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.</p>
Model Zoo: A Dataset of Diverse Populations of Neural Network Models - Fashion-MNIST
<p><strong>Abstract</strong></p> <p>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.</p> <p><strong>Dataset</strong></p> <p>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.</p> <p>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.</p> <p>For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.</p>
Model Zoo: A Dataset of Diverse Populations of Neural Network Models - MNIST
<p><strong>Abstract</strong></p> <p>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.</p> <p><strong>Dataset</strong></p> <p>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.</p> <p>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.</p> <p>For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.</p>
Event-Based Velocity Prediction for Spiking Neural Networks
<p>This dataset is intended to be used to predict the velocity based on the event pixels present in the data, given in (t, x, y, p) format alongside a ground truth velocity reading. A novel dataset using people and various objects moving in front of an RGB video camera was created. The positions of each entity and the associated times were captured with a Vicon motion tracking system. These two types of data were calibrated so that the movement in the video matched the measurements recorded by the Vicon system. The types of data collected include two different people carrying a calibrated Vicon Active Wand and moving around the room, a Lambda aerial robot with motion tracking markers that flew around the room, a box with motion tracking markers that were tossed back and forth in the air, and the same box was slid across the floor. The video recording was simulated as event camera data. Each pixel changes state independently of all the other pixels. The Open Event Camera Simulator (ESIM) from the Robotics and Perception Group at the University of Zurich and ETH Zurich is used. The event camera simulator used, ESIM, allows for accurate event simulation data. The ground truth velocities were calculated from the position and timestamps recorded by the Vicon system.</p>
nNPipe: A neural network pipeline for automated analysis of morphologically diverse catalyst systems - Resources
<p>This dataset comprises of resources required to replicate the results described in "<em>nNPipe</em>: A neural network pipeline for automated analysis of morphologically diverse catalyst systems". <em>nNPipe </em>is a deep learning based method in which two deep convolutional neural networks are used for the automated analysis of 2048x2048 HRTEM images.</p> <p>The file contains:<br> - Relevant experimental images as well as ground truth for Pd/C and Au/Ge systems.<br> - A workflow file explaining the nNPipe workflow.<br> - Mathematica 12.1 code for the generation of computational models.<br> - MATLAB code for HRTEM multislice simulations using MULTEM, as well as code required to form respective training datasets.<br> - Weights and files required for training the YOLOv5x module.<br> - Weights and files required for training the SegNet module.<br> - Mathematica 12.1 code required for reconstruction of 2048x2048 binary segmented maps of HRTEM images. </p>
Data for conductance-based simulations of "Cortical oscillations support sampling-based computations in spiking neural networks"
<p>This repository contains the full data generated by the conductance-based simulations described in: <a href="https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1009753">Cortical oscillations support sampling-based computations in spiking neural networks</a>. The code is accessible via <a href="https://doi.org/10.5281/zenodo.5512526.">this repository</a>.</p>
Landslide Susceptibility and 3-day Antecedent Rainfall generated by Artificial Neural Networks
<p>[ENGLISH]</p> <p>In this dataset, you can find:</p> <p>- Landslide susceptibility indexes for the Serra Geral geomorphic unit, from 0 (low susceptibility) to 1 (high susceptibility). Files starting in map_susc</p> <p>- 3-day Antecedent Rainfall Thresholds for rainfall-induced landslides in the Serra Geral geomorphic unit. Files starting in map_3day</p> <p>The figure tiles_location.png shows the locations of each tile within the states of Rio Grande do Sul and Santa Catarina, Brazil. Background map: OpenStreetMap contributors (2022)</p> <p>This dataset was produced within the research conducted for the PhD Thesis of Luísa Vieira Lucchese. The link to the Thesis will be added here when it is available. Reference:</p> <p>LUCCHESE, Luísa Vieira. Modelagem de Suscetibilidade e de Limiares de Precipitação para Deslizamentos de Terra utilizando métodos de Aprendizagem de Máquina. 2022. PhD Thesis (Water Resources and Environmental Sanitation) — Instituto de Pesquisas Hidráulicas, Universidade Federal do Rio Grande do Sul, Porto Alegre, 2022.</p> <p> </p> <p>[PORTUGUÊS DO BRASIL]</p> <p>Neste conjunto de dados, você encontra:</p> <p>- Índices de suscetibilidade a deslizamentos de terra para a unidade geomorfológica da Serra Geral, de 0 (baixa suscetibilidade) até 1 (alta suscetibilidade). Os arquivos têm o prefixo map_susc</p> <p>- Precipitação antecedente de 3 dias para a ocorrência de deslizamentos de terra na unidade geomorfológica da Serra Geral. Os arquivos têm o prefixo map_3day</p> <p>A figura tiles_location.png mostra a localização de cada bloco dentro dos estados do Rio Grande do Sul e de Santa Catarina. Mapa de fundo: OpenStreetMap contributors (2022)</p> <p>Este conjunto de dados é produto da Tese de Doutorado de Luísa Vieira Lucchese. O link para a Tese será adicionado aqui, quando estiver disponível. Referência:</p> <p>LUCCHESE, Luísa Vieira. Modelagem de Suscetibilidade e de Limiares de Precipitação para Deslizamentos de Terra utilizando métodos de Aprendizagem de Máquina. 2022. Tese (Doutorado em Recursos Hídricos e Saneamento Ambiental) — Instituto de Pesquisas Hidráulicas, Universidade Federal do Rio Grande Sul, Porto Alegre, 2022.</p>
CNN Wild Park - Graph Neural Networks for Learning Equivariant Representations of Neural Networks
<p>This repository contains the <strong>CNN Wild Park</strong> dataset from the paper:</p> <blockquote> <p><strong>Graph Neural Networks for Learning Equivariant Representations of Neural Networks</strong><br><a href="https://mkofinas.github.io/">Miltiadis Kofinas</a>*, <a href="https://bknyaz.github.io/">Boris Knyazev</a>, <a href="https://www.cyanogenoid.com/">Yan Zhang</a>, <a href="https://yunlu-chen.github.io/">Yunlu Chen</a>, <a href="https://gertjanburghouts.github.io/">Gertjan J. Burghouts</a>, <a href="https://egavves.com/">Efstratios Gavves</a>, <a href="https://www.ceessnoek.info/">Cees G. M. Snoek</a>, <a href="https://davzha.netlify.app/">David W. Zhang</a>*<br><em>ICLR 2024</em> (oral)<br><a href="https://arxiv.org/abs/2403.12143">https://arxiv.org/abs/2403.12143</a><br><a href="https://github.com/mkofinas/neural-graphs">https://github.com/mkofinas/neural-graphs</a><br>*Joint first and last authors</p> </blockquote> <p>We introduce a new dataset of CNNs, which we term <em>CNN Wild Park</em>.<br>The dataset consists of 117,241 checkpoints from 2,800 CNNs, trained for up to 1,000 epochs on CIFAR10.<br>The CNNs vary in the number of layers, kernel sizes, activation functions, and residual connections between arbitrary layers.</p> <p>More specifically, we construct the CNN Wild Park dataset by training 2,800 small CNNs with different architectures for 200 to 1,000 epochs on CIFAR10. We retain a checkpoint of its parameters every 10 steps and also record the test accuracy. The CNNs vary by:</p> <ul> <li>Number of layers L in [2, 3, 4, 5] (note that this does not count the input layer).</li> <li>Number of channels per layer c_l in [4, 8, 16, 32].</li> <li>Kernel size of each convolution k_l in [3, 5, 7].</li> <li>Activation functions at each layer are one of ReLU, GeLU, tanh, sigmoid, leaky ReLU, or the identity function.</li> <li>Skip connections between two layers with at least one layer in between. Each layer can have at most one incoming skip connection. We allow for skip connections even in the case when the number of channels differ, to increase the variety of architectures and ensure independence between different architectural choices. We enable this by adding the skip connection only to the min(c_n, c_m) nodes.</li> </ul> <p>We divide the dataset into train/val/test splits such that checkpoints from the same run are <strong>not</strong> contained in both the train and test splits. </p> <div> </div> <div> </div>
Data for: Tang et al., Interpretable classification of Alzheimer's disease pathologies with a convolutional neural network pipeline. bioRxiv 2018.
<p>Datasets containing 63 whole slide images (WSIs) and their segmented 256x256 pixel tiles with approximately 80,000 tile-level amyloid-β pathology expert annotations.</p> <p><strong>Paper</strong>: "Interpretable classification of Alzheimer's disease pathologies with a convolutional neural network pipeline", bioRxiv 454793; DOI: <a href="https://doi.org/10.1101/454793">https://doi.org/10.1101/454793</a>.</p> <p><strong>Details:</strong> A total of 63 WSIs for 63 unique decedent cases spanning Alzheimer’s disease (AD) to non-AD and possessing a variety of CERAD scores. WSIs comprise three datasets as follows:</p> <ol> <li><em>Development (Phases I-II)</em>. 33 WSIs used for convolutional neural network (CNN) model development (29 training, 4 validation).</li> <li><em>Hold-out (Phase III)</em>. 10 WSIs selected by an expert neuropathologist as a held-out test set to assess the generalizability of the CNN model.</li> <li><em>CERAD-like hold-out</em>. 20 blinded WSIs collected solely for use in a CERAD-like scoring comparison study.</li> </ol> <p>Datasets 1 and 2 were color-normalized and segmented to 256x256 pixel image tiles for model training set (61,370 images), validation set (8,630 images), and hold-out test set (10,873 images). Dataset 3 was color-normalized but not segmented.</p> <p>Expert labels of plaques for Dataset 1 and 2 tiles are included in corresponding CSV files.</p> <p><strong>Slide source and preparation:</strong> All samples were retrieved from archives of the University of California, Davis Alzheimer’s Disease Center Brain Bank (<a href="https://www.ucdmc.ucdavis.edu/alzheimers/">https://www.ucdmc.ucdavis.edu/alzheimers/</a>). Archival samples analyzed in this study were 5 μm formalin fixed, paraffin embedded sections of the superior and middle temporal gyrus from human brain. The tissue had been previously stained with an amyloid-β antibody (4G8, recognizing residues 17-24, BioLegend, formerly Covance) that were first pretreated with formic acid to rid samples of endogenous protein. All slides were digitized using an Aperio AT2 up to 40x magnification.</p> <p><strong>Code:</strong> Please visit <a href="https://github.com/keiserlab/plaquebox-paper">https://github.com/keiserlab/plaquebox-paper</a></p> <p> </p>
Sensitivity Datasets - Leveraging Implicit Knowledge in Neural Networks for Functional Dissection and Engineering of Proteins
<p><strong>Leveraging Implicit Knowledge in Neural Networks for Functional Dissection and Engineering of Proteins</strong></p> <p>The Sensitivity datasets cover more than 800 proteins and are structured as follows. The sensitivity values are the mean of four DeeProtein replicates.</p> <p>It is uploaded as tar.gz. and contains one directory.</p> <p>File names contain the PDB<sup>1</sup> identifier and the respective chain identifier. </p> <p>The sequences and secondary structure information were downloaded from the RCSB Protein Databank and are available here: <a href="https://cdn.rcsb.org/etl/kabschSander/ss_dis.txt.gz">https://cdn.rcsb.org/etl/kabschSander/ss_dis.txt.gz</a> This URL can be found with some explanation at <a href="http://www.rcsb.org/pdb/static.do?p=download/http/index.html">http://www.rcsb.org/pdb/static.do?p=download/http/index.html</a></p> <p>The secondary structure annotation relies on the DSSP Algorithm by Kabsch and Sander<sup>2</sup>.</p> <p> </p> <p><strong>The files are tab-separated and contain the following columns:</strong></p> <ul> <li><strong>Pos</strong> Position in the sequence, starting from zero</li> <li><strong>AA</strong> Amino acid in that position</li> <li><strong>sec</strong> Secondary structure as annotated in the RCSB Protein Databank</li> <li><strong>dis</strong> if a region has not been experimentally observed (sometimes explains mismatches with crystal structures)</li> <li><strong>GO:_______</strong> Sensitivity for the GO term</li> </ul> <p><strong>References</strong></p> <ol> <li>The Protein Data Bank H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig, I.N. Shindyalov, P.E. Bourne (2000) Nucleic Acids Research, 28: 235-242. doi:10.1093/nar/28.1.235</li> <li>Kabsch, W. & Sander, C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22, 2577-2637, doi:10.1002/bip.360221211 (1983).</li> </ol>
Data for: "Comprehensive sampling of coverage effects in catalysis by leveraging generalization in neural network models"
<p>This repository contains the raw data to reproduce the paper: "Comprehensive sampling of coverage effects in catalysis by leveraging generalization in neural network models". Within the .tar.gz file, you will find the directory structure described above.</p> <h2>Directory Structure</h2> <h3>`data`</h3> <p>Contains the data to reproduce all figures in the manuscript. Used primarily by the Jupyter Notebooks that plot the data from the paper.</p> <h3>`eval`</h3> <p>Contains the predicted energies according to a MACE model for the following systems and facets:<br>- covsplit (100, 111, 211, 331, 410, 711): The NN model is trained on low-coverage structures and tested on high-coverage structures for a single facet<br>- evencov (100, 111, 211, 331, 410, 711): The NN is trained on even coverages and tested on odd coverages for a single facet<br>- facet (100, 111, 211, 331, 410, 711): the NN is trained on the facet indicated by the folder name (e.g., facet-100 means that the model was trained on Cu(100)) and tested on all of the other facets.<br>- full: the model was trained on all facets and all coverages<br>- slopes (various versions and configurations): the models were trained with different body-order correlation (v) for the Cu(711) facet and tested only on the Cu(711) facet<br>- Rh111: Energies for the Rh(111) + CHOH + CO systems.</p> <h3>`mcmc`</h3> <p>Contains the data for MCMC (Markov Chain Monte Carlo) evaluations for two systems: Cu and Rh<br>- copper-mcmc-public.tar.gz<br>- rhodium-mcmc-public.tar.gz</p> <h3>`models`</h3> <p>Contains the weights and parameters of the best-performing MACE models trained in this work, as selected by the validation loss:</p> <p>File formats: `.model` and `_swa.model` relate to the first-stage of training and the second-stage of training.</p> <h3>`pyscripts`</h3> <p>Python scripts to perform the MCMC sampling given the custom configuration file `sample_cfg.json`.</p> <h3>`scripts`</h3> <p>Shell scripts for evaluation and training the MACE models, along with the hyperparameters used in doing so.</p> <p>- Evaluation scripts (eval-*.sh)<br>- Training scripts (train-*.sh)</p> <h3>`train`</h3> <p>Training, validation, and testing data for all Cu and Rh facets in this work, according to the naming scheme described above.</p> <p>- Rh111<br>- covsplit<br>- evencov<br>- facet<br>- full<br>- slopes</p>
Neural Network and objective analysis reconstruction of 3D Mediterranean physical fields from surface satellite and in situ observations at 1/24 deg
<p>Daily Mediterranean 3D fields of temperature, salinity and geostrophic current at 1/24° of resolution, up to 150m-depth and from 2016 to mid 2022, obtained through a 3 steps approach: (1) Temperature and salinity 3D fields have been first estimated by a machine learning approach by using mediterranean reanalysis outputs (https://doi.org/10.25423/CMCC/MEDSEA_MULTIYEAR_PHY_006_004_E3R) together with satellite observations, (2) a combination of this first step with in situ observations through an Optimal interpolation to remove part of large scale biases, (3) the computation of geostrophic currents using the thermal wind equation. This work has been funded by the European Space Agency through the 4DMED-SEA project [ESA contract No. 4000141547/23/I-DT].</p>
Predicting continuous ground reaction forces from accelerometers during uphill and downhill running: A recurrent neural network solution
<p>Data and model files supporting the manuscript: </p> <p>Predicting continuous ground reaction forces from accelerometers during uphill and downhill running: A recurrent neural network solution.</p> <p>Repository: https://github.com/alcantarar/Recurrent_GRF_Prediction</p>
Dataset for "A Bayesian neural network predicts the dissolution of compact planetary systems"
<p>The dataset used for training and evaluating the models in the paper "A Bayesian neural network predicts the dissolution of compact planetary systems": https://arxiv.org/abs/2101.04117. </p> <p>The code for working with this dataset, and other links, can be found at: https://github.com/MilesCranmer/bnn_chaos_model.</p>
Photometric Redshifts for Cosmology: Improving accuracy and uncertainty estimates using Bayesian Neural Networks
<p><strong>This data consists of 286,401 with broad-band g,r,i,z,y photometry from the HSC DR2 survey and spectroscopic redshifts. The majority of galaxies in our sample lies between redshift of 0.01 and 2.5</strong></p>
Prediction of inter-chain distance maps of protein complexes with 2D attention-based deep neural networks
<p>Residue-residue distance information is useful for predicting tertiary structures of protein monomers or quaternary structures of protein complexes. Many deep learning methods have been developed to predict intra-chain residue-residue distances of monomers accurately, but few methods can accurately predict inter-chain residue-residue distances of complexes. We develop a deep learning method CDPred (i.e., Complex Distance Prediction) based on the 2D attention-powered residual network to address the gap. Tested on two homodimer datasets, CDPred achieves the precision of 60.94% and 42.93% for top L/5 inter-chain contact predictions (L: length of the monomer in homodimer), respectively, substantially higher than DeepHomo’s 37.40% and 23.08% and GLINTER’s 48.09% and 36.74%. Tested on the two heterodimer datasets, the top Ls/5 inter-chain contact prediction precision (Ls: length of the shorter monomer in heterodimer) of CDPred is 47.59% and 22.87% respectively, surpassing GLINTER’s 23.24% and 13.49%. Moreover, the prediction of CDPred is complementary with that of AlphaFold2-multimer.</p>
Database of small molecule X-ray absorption spectra, featurized structures, and neural network ensembles
<p>Companion data for arXiv preprint <em>Uncertainty-aware predictions of molecular X-ray absorption spectra using neural network ensembles</em> (<a href="https://arxiv.org/abs/2210.00336">https://arxiv.org/abs/2210.00336</a>), by Animesh Ghose, Mikhail Segal, Fanchen Meng, Zhu Liang, Mark S. Hybertsen, Xiaohui Qu, Eli Stavitski, Shinjae Yoo, Deyu Lu & Matthew R. Carbone.</p> <p><strong>Included</strong></p> <ul> <li>*-XANES-*.tar.bz2: raw input/output files for all molecular simulations used in the work. These inputs and outputs correspond to the structural data in the QM9 dataset.</li> <li>ml_ready.tar.bz2: machine learning-ready data (featurized spectra). Used as input to the neural network ensembles.</li> <li>XANES-220712-ACSF-*.tar.bz2: neural network ensembles used in this work.</li> </ul> <p><strong>Notes</strong></p> <ul> <li>The FEFF9 code [J. J. Rehr, J. J. Kas, F. D. Vila, M. P. Prange, and K. Jorissen, <em>Phys. Chem. Chem. Phys.</em> <strong>12</strong>, 5503 (2010)] was used to generate all X-ray absorption near-edge structure (XANES) spectra.</li> <li>All molecular structures were sourced from the QM9 database [R. Ramakrishnan, P. O. Dral, M. Rupp, and O. A. Von Lilienfeld, <em>Sci. Data</em> <strong>1</strong>, 1 (2014)].</li> </ul> <p><strong>Funding</strong></p> <p>This research is based upon work supported by the U.S. Department of Energy, Office of Science, Office Basic Energy Sciences, under Award Number FWP PS-030. This research also used theory and computational resources of the Center for Functional Nanomaterials, which is a U.S. Department of Energy Office of Science User Facility, and the Scientific Data and Computing Center, a component of the Computational Science Initiative, at Brookhaven National Laboratory under Contract No. DE-SC0012704.</p>
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.