Find research datasets worth reusing

opencc-by-4.0Apr 2024View details →

Supplementary Dataset for the Paper: "Parameter-Adaptive Approximate MPC: Tuning Neural-Network Controllers without Re-Training"

Supplementary dataset for paper Hose, Henrik, Alexander Gräfe, and Sebastian Trimpe. "Parameter-Adaptive Approximate MPC: Tuning Neural-Network Controllers without Re-Training." arXiv preprint arXiv:2404.05835 (2024). The code to use this dataset is publicly available at <a href="https://github.com/hshose/Adaptive-AMPC-Cartpole">https://github.com/hshose/Adaptive-AMPC-Cartpole</a> The dataset contains training and testing data to train an NN controller for a standart cartpole system. For each system, there are initial conditions as comma separated value in the `x0.txt` file, the MPC input trajectory in the `U.txt` file and the corresponding predicted state sequence in the `X.txt` file. Additionally, sensitivities (i.e. gradients dU/dtheta with respect to some system parameters theta) are provided in a file called `J.txt`.

opencc-by-nc-4.0May 2024View details →

Exploring the Search Space of Neural Network Combinations obtained with Efficient Model Stitching - Results Data

Open the record for dataset details and reuse information.

Predictive Modeling of Bearing Degradation: LSTM Neural Networks for Uncertainty Quantification

These MATLAB codes are part of a research project focused on predicting bearing degradation through vibration measurements. The codes implement LSTM (Long Short-Term Memory) neural network models trained under different objectives, including uncertainty quantification and RMSE (Root Mean Square Error) minimization. The objective of the research is to compare the performance of these models in predicting bearing health and assessing the associated uncertainty. Note: The current codes are under embargo access as the corresponding paper has been submitted to the ESCA 11 conference. The codes will be made openly accessible upon acceptance of the paper and during the presentation dates. Please cite our paper when using these codes.

MATLAB Implementation for Wind Turbine Prognosis Using Uncertainty Bayesian-Optimized Lightweight Neural Network

These MATLAB codes accompany the paper titled "---," currently submitted to the 11th International Electronic Conference on Sensors and Applications (ECSA-11). The paper presents a novel approach to wind turbine prognosis for maintenance purposes using the Uncertainty Bayesian-Optimized Extreme Learning Machine (UBO-ELM) algorithm. The codes provided here implement the methodology described in the paper, including data preprocessing, model training and evaluation, uncertainty quantification, and visualization of results. These codes are intended for researchers and practitioners in the field of wind energy systems and predictive maintenance. Please note that the paper is currently under review at ECSA-11. Once the paper is approved and the embargo is lifted, these codes will be accessible openly. Users are kindly requested to cite our paper when utilizing these codes for their research.

Dataset of paper "Neural Network Modeling of Black Box Controls for Internal Combustion Engine Calibration"

These are the two training datasets for the Exhaust Temperature Model (ETM) and the Three-Way Catalyst Control Lambda Set Point (TWCC) Model, used in Black-Box Modeling, such as through Neural Networks. The datasets for each black box model are stored in XLSX files. Each dataset file contains ECU channel inputs, related Map parameters, and output. The ECU channel inputs and Map parameters are collected with Sobol methods. The output is generated by Hardware-in-the-Loop (HiL) corresponding to the related ECU channel inputs and Map parameters. The ETM model has four effective Maps, hence there are 32 corresponding shape-based algorithm parameters. Its output is the exhaust temperature. Its ECU channel inputs are as follows: CH_1:       Intake valve cut-off condition CH_2:       Thrust cut-off condition CH_3:       Cylinder equalization control work cycle injection enabled CH_4:       Ignition angle efficiency CH_5:       Relative level of reduction CH_6:       Lean engine lambda for cylinders on exhaust bank with lambda split CH_7:       Rich engine lambda for cylinders on exhaust bank with lambda split CH_8:       Exhaust gas mass flow CH_9:       Engine speed CH_10:     Relative air filling CH_11:     Material temperature CH_12:     Engine temperature   The TWCC model has one Map, hence there are 8 corresponding shape-based algorithm parameters. Its output is lambda set point. Its ECU channel inputs are as follows: CH_1:       Exhaust mass flow CH_2:       Trigger bit for inhibiting the purge CH_3:       Trigger bit for a rich mixture CH_4:       Trigger bit for empty catalyst CH_5:       Trigger bit for catalyst reset CH_6:       Oxygen storage level CH_7:       Desired air-fuel ratio

Windy events detection in big bioacoustics datasets using a pre-trained Convolutional Neural Network

This repository includes the code and all relevant files used throughout our study. These encompass everything from the initial sheets of the whole acoustic dataset utilised for selecting the annotated dataset to the notebook (.ipynb) and the recordings employed in training the model. The .wav file are compressed in the file: <a href="../api/records/11220741/draft/files/wind-noise-detection-main.7z/content" target="_blank" rel="noopener noreferrer">wind-noise-detection-main.7z</a>/data/208_file_recordings_paper_wind.tar

Synthetic dataset and prediction files for the paper "Denoising of Geodetic Time Series Using Spatiotemporal Graph Neural Networks: Application to Slow Slip Event Extraction", by Costantino et al. (2024)

Synthetic database used for training and evaluation of SSEdenoiser

Data for "Investigating molecular transport in the human brain from MRI with physics-informed neural networks"

Data analyzed in Zapf et al. <a href="https://www.nature.com/articles/s41598-022-19157-w">Investigating molecular transport in the human brain from MRI with physics-informed neural networks</a> (Scientific Reports 2022). The data consists of CSF tracer concentrations in the brain subregions analyzed in the article. The data was pre-processed as described in S1.1 in the supplementarty materials.  In Python, load the data as numpy arrays using the nibabel package as <pre><code>import nibabel data = nibabel.load("068/concentrations/24h.mgz").get_fdata() domain_mask = nibabel.load("068/masks/roi.mgz").get_fdata().astype(bool) And view slices of the data as:</code></pre> <div> <div>plt.figure()</div> <div>plt.imshow(np.take(data, 150, 0), vmax=0.1)</div> <div>plt.figure()</div> <div>plt.imshow(np.take(data, 100, 1), vmax=0.1)</div> <div>plt.figure()</div> <div>plt.imshow(np.take(data, 100, 2), vmax=0.1)</div> <div>plt.show()</div> </div> <pre> </pre>

Supporting data for "Neural Network-Based Interatomic Potential for the Study of Thermal and Mechanical Properties of Siliceous Zeolites"

Open the record for dataset details and reuse information.

Knowledge Graph Neural Network with Spatial-Aware Capsule for Drug-Drug Interaction Prediction

Open the record for dataset details and reuse information.

Dataset for: Trait-mediated speciation and human-driven extinctions in proboscideans revealed by unsupervised Bayesian neural networks

This repository contains the output of all birth-death neural network simulations and the empirical analyses of Proboscidean diversification. For details see the Readme of the individual directories.

PLANT SPECIES RECOGNITION USING LEAF IMAGES AND CONVOLUTIONAL NEURAL NETWORKS (CAAR dataset, version 1)

The CAAR dataset contains leaf images from plants obtained from the Arboreal Collection at Augusto Ribas Agricultural College (CAAR/UEPG). This plant collection is situated in Augusto Ribas College, located at the Ponta Grossa State University, Ponta Grossa, Paraná, Brasil. The images were taken using a smartphone camera with a resolution of 1659 x 2658 pixels and 24 bits of color depth. For each plant, images samples were collected using a white paper sheet background, varying the leaf orientation. The number of plant species used to build the dataset is equal to 35. Data augmentation, using rotation and zoom, were used to increase the data size from 730 to 1986 images in this dataset.

Dataset of miRNA-Disease Relations Extracted from Textual Data using Transformer-based Neural Networks

Supplementary Data.

Computation noise promotes zero-shot adaptation to uncertainty during decision-making in artificial neural networks

This dataset contains the behavioral choice data obtained from N = 230 participants that played a two-armed bandit task (139 females, age: 34 +/- 10 years) in partial and complete feedback conditions, as described in (Findling, Skvortsova et al., 2019, Nature Neuroscience, https://doi.org/10.1038/s41593-019-0518-9). <div> <div> <div> The experiment was performed on the Prolific platform (prolific.co) and the research was carried out following the principles and guidelines for experiments including human participants provided in the declaration of Helsinki and approved by the relevant authorities (Inserm Ethical Review Committee, IRB #00003888). All participants provided written informed consent prior to their inclusion. </div> </div> </div>

Achieving Peak Performance: Unveiling the Best Hyperparameter Combination for Accurate Skin Disorders Pose Classification with Convolutional Neural Network

This material has presented on 2nd International Conference on Advance Research in Social and Economic Science in October 25, 2023.

Data from: Using artificial neural networks and citizen science data to assess jellyfish presence along coastal areas

General Information This dataset was used in the study titled "Using artificial neural networks and citizen science data to assess jellyfish presence along coastal areas". The study employs citizen science data collected from the Infomedusa application to assess the presence of jellyfish on beaches along the Andalusian coast, along with environmental data to analyze the factors influencing jellyfish distribution. The study aims to employ machine learning techniques, specifically a Multi-Layer Perceptron (MLP) neural network, to classify user comments on the presence or absence of jellyfish and analyze how environmental factors such as sea surface temperature, wind direction, and wind speed influence jellyfish distribution. Dataset Columns <ul> <li>Fecha: Timestamp of the comment made by the user in the Infomedusa application.</li> <li>Municipio: Name of the municipality where the beach mentioned in the comment is located.</li> <li>Jellyfish: Binary variable indicating the presence (1) or absence (0) of jellyfish according to the user’s comment.</li> <li>Comunidad: Autonomous community to which the municipality belongs.</li> <li>Provincia: Province to which the municipality belongs.</li> <li>Latitud: Geographical latitude of the municipality where the comment was made.</li> <li>Longitud: Geographical longitude of the municipality where the comment was made.</li> <li>Set: Set of grouped beaches for geographical analysis. Each set includes beaches close to each other and the nearest weather station.</li> <li>Month: Month when the comment was made.</li> <li>Longitud_sea: Longitude of the nearest point in the sea for which environmental data was available.</li> <li>Latitud_sea: Latitude of the nearest point in the sea for which environmental data was available.</li> <li>SST: Sea Surface Temperature at the nearest point in the sea to the municipality, obtained from the Copernicus Marine Environment Monitoring Service.</li> <li>Wind_dir: Wind direction measured at the weather station closest to the municipality, provided by the Spanish Meteorological Agency (AEMET).</li> <li>Wind_speed: Wind speed measured at the weather station closest to the municipality, provided by the AEMET.</li> </ul> Data Sources <ul> <li>Infomedusa APP: Application developed by the Provincial Council of Malaga and Aula del Mar of Malaga to monitor the presence of jellyfish through citizen participation.</li> <li>Copernicus Marine Environment Monitoring Service (CMEMS): Provides data on sea surface temperature with an hourly temporal resolution and a spatial resolution of 0.0625° x 0.0625°.</li> <li>Agencia Estatal de Meteorología (AEMET): Provides daily data on wind direction and speed.</li> </ul>

Convolutional Neural Networks and their Activations: An Exploratory Case Study on Mounded Settlements

This data set represents the output of our paper "Convolutional Neural Networks and their Activations: An Exploratory Case Study on Mounded Settlements". It comprises png images of 400x800 pixels. Half of the image represents a section of CORONA imagery where a site (files named Site_.png) or no site (files named NonSite_.png) is present. The other half of the image contains the activation map produced by various explainability techniques.

Train and Test Dataset for a Neural Network for the task of RSA Segmentation in the context of a Master thesis 2024

The data corresponds to the work of the master thesis "3D Segmentation of Plant Roots from MRI Images for Enhanced Automated Root Tracing using Deep Neural Networks" by Daniel Weißen from the 24/07/2024 train_data.tar - contains the training and validation data used for the DNN   - data_gmm contains the soil noise data based on a Gaussian Mixture Model   - data_FS contains the soil noise data based on the Fourier Synthesis approach described in the thesis test_data.tar - contains the test data which is manually labeled segmentations.tar - contains the segmentations of the trained SwinUNETR-V2+ on the real test MRIs   - fs_model_output: contains the Fourier Synthesis trained NNs segmentation   - fmm_model_output: contains the GMM trained NNs segmentation   - threshold_output: contains the Dice optimized threshold labeling FS_checkpoint.ckpt, GMM_checkpoint.ckpt - contains the checkpoint for the trained SwinUNETR-V2+ on the Fourier synthesis, and GMM soil noise respectively

Modeling fluid flow in ship systems for controller tuning using an artificial neural network

Dataset used to develop ANN NARX models