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.
2,025
datasets available to search
ShareScore release 0.9.0
Dataset results
2,025 results for “AIS”
Perceptions and Knowledge regarding the application of Artificial Intelligence (AI) technologies in healthcare system; A cross-sectional Study involving Medical Students.
<p>This is the data set of a cross-sectional study that was carried out on medical students at Rawalpindi medical university, Pakistan involving 211 participants.</p>
Interactive ethical AI quiz
<p>Watch this demo video for a step-by-step explanation of how to use the <a href="https://interactive-ai-ethics-quiz.herokuapp.com/">Interactive Ethical AI quiz.</a></p>
A Case Study on AI Engineering Practices: Developing an Autonomous Stock Trading System
<p>The field note data related to a case study on AI engineering practices. During the case study, an autonomous stock trading system was developed, while 10 AI engineering practices were applied. We used field notes to document the application of each practice. Additionally, we documented any challenges we experienced.</p> <ul> <li>field-notes-practices.csv: the documented experiences with the AI engineering practices containing the name of the practice, the perceived effectiveness of the practice on an ordinal scale (--, -, 0, +, ++), a description of how it was applied and its effects, and the date when it was applied (in DD.MM.YYYY format).</li> <li>field-notes-challenges.csv: the documented challenges experiences during the case study containing the date (in DD.MM.YYYY format), a description of the challenge, the source that made us became aware of the challenge, the solution we used to overcome the challenge, and a description of the frequency of occurrence.</li> </ul>
Design Patterns for AI-based Systems: A Multivocal Literature Review and Pattern Repository
<p>The data for a multivocal literature review on design patterns for AI-based systems.</p> <ul> <li>mlr-search-and-selection.xlsx: the results from the queried databases and search engines, the inclusion/exclusion process, and the backward and forward snowballing results</li> <li>mlr-results.xlsx: the final set of selected resources, the patterns extracted from them, and some analysis</li> <li>query-strings-google-and-google-scholar.txt: the individual terms of the search query (broken up for Google Scholar and Google Search)</li> </ul>
Datasets used in the paper "The Face of Deception: The Impact of AI-Generated Photos on Malicious Social Bots"
<p>Datasets used in the paper "The Face of Deception: The Impact of AI-Generated Photos on Malicious Social Bots"</p> <p>We changed the datasets' titles and omitted authors' names for the blind review process. After the review, we will upload it to GitHub in an unanonymised form.</p> <p>Check README.md for details.</p>
CHclim25 - aridity index (AI)
<p>Aridity index is calculated as in the <a href="https://cgiarcsi.community/data/global-aridity-and-pet-database/">CGIAR/CSI dataset</a> from CHclim25 sum of precipitation (Prec) and CHclim25 potential evapotranspiration (etp). Yearly individual years, current average (1981-2010) and future average (2020-2049, 2045-2074, and 2070-2099)<strong> </strong>layers can be downloaded from separate zip files. </p> <p>Future layers are based on the transient daily time series of gridded climate scenarios of temperature at 0.02°D (~2.2 km) provided by the <a href="https://www.nccs.admin.ch/nccs/en/home/climate-change-and-impacts/swiss-climate-change-scenarios/ch2018---climate-scenarios-for-switzerland.html">CH2018 initiative</a>. We calculated future climatic layers for 4 GCMs (HADGEM, ECEARTH, MPIESM, and IPSL), 3 time slices (2020-2049, 2045-2074, and 2070-2099) and 3 representative concentration pathways (RCP 2.6, 4.5 and 8.5)</p> <p>The layer files are stored in compressed GeoTIFF format with the “deflate” algorithm with option “predictor2” from the GDAL. This format has a high compression ratio but allows direct import in most GIS softwares. All the maps are projected in the Swiss coordinate system CH 1903+ LV95 (epsg:2056) with a resolution of 25x25m using the extent of the digital height model DHM25 of the Swiss office for topography (swisstopo).</p>
AI in the US media between the summer of 2021 and 2022
<p>SPSS file of media analysis on AI for between summer of 2021 and summer of 2022. The analyis looks at the hype and disappointment centeredness of the discourse, the costs and benefits presented in the outlets amongst others. </p>
Performance monitoring leveraging advanced AI technique with CNN
<p>The main goal of this project is to study and develop a reliable nondestructive testing (NDT)-based structural performance prediction model framework leveraging the advanced machine learning convolutional neural network (CNN) technique and rapid crack evaluation system. There are two steps of application CNN technique in this project: 1) the first step is to identify delamination, noise, and the unexpected signal produced by the existing damage identification algorithm to improve the accuracy of NDT results. The input image or training data of NDT data for CNN is comprehensively studied with several features, such as the duration of the signal, the starting time of the signal, the resolution of images, and the number of images. 2) The second step is to study damage prediction with four different stress levels. The FE model is used to simulate structural performance with different delamination conditions. Moreover, except for field test results, the artificial delamination model is created. We performed numerous finite element (FE) simulation to create inputs for CNN for damage detection. The result shows improved NDT results, and CNN can achieve structural performance prediction. We performed six tasks based on these objectives: Task 1. literature review; Task 2. Collect data from bridges; Task 3. perform filed test NDT results; Task 4. Develop FE model based on field test results; Task 5. development of a machine learning model for damage prediction.</p>
Explainable AI for Retinoblastoma Diagnosis: Interpreting Deep Learning Models with LIME and SHAP
<p>Retinoblastoma is a rare and aggressive form of childhood eye cancer that requires prompt diagnosis and treatment to prevent vision loss and even death. Deep learning models have shown promising results in detecting retinoblastoma from fundus images, but their decision-making process is often considered a "black box" that lacks transparency and interpretability. In this project, we explore the use of LIME and SHAP, two popular explainable AI techniques, to generate local and global explanations for a deep learning model based on InceptionV3 architecture trained on retinoblastoma and non-retinoblastoma fundus images. We collected and labeled a dataset of 400 retinoblastoma and 400 non-retinoblastoma images, split it into training, validation, and test sets, and trained the model using transfer learning from the pre-trained InceptionV3 model. We then applied LIME and SHAP to generate explanations for the model's predictions on the validation and test sets. Our results demonstrate that LIME and SHAP can effectively identify the regions and features in the input images that contribute the most to the model's predictions, providing valuable insights into the decision-making process of the deep learning model. In addition, the use of InceptionV3 architecture with spatial attention mechanism achieved high accuracy of 97\% on the test set, indicating the potential of combining deep learning and explainable AI for improving retinoblastoma diagnosis and treatment.</p>
Replication Package for the paper "AI-based Fault-proneness Metrics for Source Code Changes"
<p>This is the replication package for the paper "<em>AI-based Fault-proneness Metrics for Source Code Changes</em>", submitted at the <em>IWSM-Mensura '23 </em>conference.</p> <p>The archive is a <em>Docker </em>image file with a fully setup and working environment to re-execute the experiments involved in the manuscript. We pre-loaded all libraries and codeBERT models to ease the replication process and avoid compatibility issues, as the environment cannot be easily managed using <em>Dockerfile</em>s.</p> <p>To run the image, a <em>Docker</em> installation is needed. Once downloaded, from the command line type:</p> <pre><code>docker load -i </path/to/downloaded/ai-proneness-replication.tar></code></pre> <p>After the loading process, you can run the container by typing:</p> <pre><code>docker run -it mensura/ai-proneness-replication:1.0</code></pre> <p>All the source code and the dataset to re-execute the experiment is located into the <em>/Replication</em> folder. The folder contains the results of our experimentation in CSV and MS Excel format, along with the following subdirectories:</p> <ul> <li><em>dataset</em>: a replication of the used dataset. The file <em>dataset.csv</em> gives information on all the entries, while the <em>code </em>folder contains a subdirectory for each sample, named by its id. In the folder, the file <em>old.txt </em>and<em> </em><em>new.txt </em>refers to the older and newer version of the method, respectively; <em>gitdiff.txt </em>stores the raw <em>git-diff</em> command output, while <em>diff.html</em> stores a more human-readable version of the differences.</li> <li><em>ai-fault-proneness-tk-replication</em>: the Java code used to apply Tree Kernel techniques on the dataset (we used JDK-11, embedded within the container). To build and execute the package, refer to the file <em>README.md</em> in the folder. For convenience, we also provided an executable JAR file <em>ai-fault-proneness-tk-replication-1.0-jar-with-dependencies.jar </em>that can be run directly and saves the output in a CSV file in the <em>results</em> folder of the replication package.</li> <li><em>code-embeddings-and-analysis</em>: python scripts to execute the <em>codeBERT</em>-based approaches and to extract the <em>diff</em> statistics. To execute all the steps, a convenience shell script <em>execute.sh</em> has been pre-loaded and can be executed to automatize all the process.</li> </ul>
Digitalizing Europe: The promising impact of AI on Car Safety Development
<p>Video through which the ESR09 Giada Colella applied to the MSCA Science-Policy Pitch Competition and Awards.</p>
Performance Model, Deliverable D2.4 Final Release and Evaluation of the AI-SPRINT Design Tools.
<p>This repository includes the code and results of analysis for predicting unseen layers of the same network and predicting new unseen networks reported in Deliverable D2.4 Final Release and Evaluation of the AI-SPRINT Design Tools<strong>. </strong>The folder is organised as follows:</p> <p>- **pareto** contains the results of the profiling of the networks trained and validated on CIFAR-10</p> <p>- **pareto_onnx** contains the results of the profiling of the networks trained and validated on MNIST</p> <p>- The folders "**network_to_network**" contain the models trained using aMLLibrary for the analysis on unseen networks done using Eduard's data. </p> <p>- **notebook-B5.ipynb** contains the code to perform the analysis on B and C networks</p> <p>- **nextLayerPrediction.ipynb** contains the code to perform the prediction of unseen layers of the same network</p> <p>- **allLayerPrediction.ipynb** contains the code for the analysis on unseen networks</p>
Replication package of "How Do Deep Learning Faults Affect AI-Enabled Cyber-Physical Systems in Operation? A Preliminary Study Based on DeepCrime Mutation Operators"
<p>Cyber-Physical Systems (CPSs) combine digital cyber technologies with physical processes. As in any other software system, in the case of CPSs, the use of Artificial Intelligence (AI) techniques in general, and Deep Neural Networks (DNNs) in particular, is contantly increasing. While recent studies have considerably advanced the field of testing AI-enabled systems, it has not yet been investigated how different Deep Learning (DL) bugs affect AI-enabled CPSs in operation. This work-in-progress paper presents a preliminary evaluation on how such bugs can affect CPSs in operation by using a mobile robot as a case study system. For that, we generated DL mutants by using operators proposed by Humbatova et al., which are operators based on real-world DL faults. Our preliminary investigation suggests that such bugs are more difficult to detect when they are deployed in operation rather than when testing their DNN in an off-line setup, which contrast with related studies.</p> <p> </p> <p>This repository provides the replication data employed in our study.</p>
Replication package for: "Corrupted by Algorithms? How AI-generated and Human-written Advice Shape (Dis)honesty"
<p>Package to the following paper:</p> <p>Leib, M; Köbis, N; Rilke, R M; Hagens, M; Irlenbusch, B (2023) Corrupted by Algorithms? How AI-generated and Human-written Advice Shape (Dis)honesty</p>
Data from: Two complementary AI approaches for predicting UMLS semantic group assignment: heuristic reasoning and deep learning
<p><strong>Objective</strong>: Use heuristic, deep learning (DL), and hybrid AI methods to predict semantic group (SG) assignments for new UMLS Metathesaurus atoms, with target accuracy ≥ 95%.</p> <p><strong>Materials and Methods</strong>: We used train-test datasets from successive 2020AA-2022AB UMLS Metathesaurus releases. Our heuristic "waterfall" approach employed a sequence of seven different SG prediction methods. Atoms not qualifying for a method were passed on to the next method. The DL approach generated BioWordVec and SapBERT embeddings for atom names, BioWordVec embeddings for source vocabulary names, and BioWordVec embeddings for atom names of the second-to-top nodes of an atom's source hierarchy. We fed a concatenation of the four embeddings into a fully connected multi-layer neural network with an output layer of 15 nodes (one for each SG). Both methods were capable of estimating the probability that their predicted SG for an atom would be correct. We developed two hybrid SG prediction methods combining the strengths of heuristic and DL methods.</p> <p><strong>Results</strong>: The heuristic waterfall approach accurately predicted 94.3% of SGs for 1,563,692 new unseen atoms. The DL accuracy on the same dataset was also 94.3%. The hybrid approaches achieved an average accuracy of 96.5%.</p> <p><strong>Conclusion</strong>: Our study demonstrated that AI methods can predict SG assignments for new UMLS atoms with sufficient accuracy to be potentially useful as an intermediate step in the time-consuming task of assigning new atoms to UMLS concepts (CUIs). We showed that for SG prediction, combining heuristic methods and DL methods can produce better results than either alone.</p>
Primo giorno di interviste ai vincitori nelle varie categorie del Premio Nazionale di Filosofia, edizione 2023
<p>Dopo la premiazione avvenuta il 18 giugno 2023 a Certaldo (FI), c'è un evento/convegno online in cui i vincitori fanno da relatori parlando delle loro opere premiate, modello intervista. Questo primo giorno su tre giorni previsti di interviste online, ha la data del 20 luglio 2023 a un mese esatto dalla premiazione in presenza.</p>
Loghub: A Large Collection of System Log Datasets for AI-driven Log Analytics
<p>Loghub datasets published in ISSRE 2023. See more details at https://github.com/logpai/loghub</p> <p>License: The datasets are freely available for research or academic work, subject to the following condition: For any usage or distribution of the loghub datasets, please refer to the loghub repository URL (https://github.com/logpai/loghub) and cite the loghub paper (Loghub: A Large Collection of System Log Datasets for AI-driven Log Analytics) where applicable.</p> <p>The above license notice shall be included in all copies of the datasets.</p>
Fox8-23 AI-powered social bot dataset
<p>We release the fox8-23 dataset, which contains the recent tweets from 1140 AI-powered social bots and 1140 human accounts collected from Twitter.</p> <p>See the GitHub repo https://github.com/osome-iu/AIBot_fox8 for details.</p>
Terzo e ultimo giorno di interviste ai vincitori del Premio Nazionale di Filosofia, 31 luglio 2023
<p>Terzo e ultimo giorno di interviste ai vincitori del Premio Nazionale di Filosofia, edizione 2023.</p> <p>Le interviste si sono svolte il 31 luglio 2023.</p> <p>Le testimonianze e le sensazioni dei vincitori e delle vincitrici in veste di relatori e di relatrici secondo un modello comunicativo a semi-intervista.</p>
Terzo e ultimo appuntamento con le interviste ai vincitori del Premio Nazionale di filosofia, edizione 2023
<p>Il terzo e ultimo appuntamento con le interviste ai vincitori e alle vincitrici del Premio Nazionale di filosofia - edizione 2023 - si è svolto nel tardo pomeriggio del 31 luglio 2023, ricco delle relazioni delle vincitrici e dei vincitori effettuate modello intervista.</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.