Find research datasets worth reusing

Do AI assistants help students write formal specifications? A study with ChatGPT and the B-Method

<p>Replication package of the paper:  "Do AI assistants help students write formal specifications? A study with ChatGPT and the B-Method."  <br>Submitted to the 37th edition of the IEEE Conference on Software Engineering Education and Training (CSEE&T), co-located with ICSE 2025. </p>

Using AI Algorithms for Predictive Analysis in Personalized Medicine

<p><strong><span>This study explored the factors influencing patients' willingness to adopt AI-powered personalized medicine. This research found the problems. Integrating AI and personalized medicine has the potential to revolutionize healthcare. However, public trust in AI for healthcare applications remains a challenge. This research examines the factors determining people's views toward using artificial intelligence for predictive analytics in personalized medicine. A cross-sectional design was employed through a survey distributed via Google Forms in April 2024 using purposive sampling. The target respondents included residents of the Jabodetabek area (Jakarta, Bogor, Depok, Tangerang, Bekasi- cities in Indonesia) with prior experience seeking medical consultation or checkups. A total of 267 responses were collected after removing outliers. The study used a Partial Least Squares Structural Equation Modeling (PLS-SEM) approach to analyze the data. </span></strong><strong><span>The study considered six independent variables: AI knowledge, trust in AI, attitude towards data privacy, personalized medicine expectations, personalized medicine understanding, and perceived risk of discrimination in AI. The dependent variable was the intention to use AI in personalized medicine. It found five of six hypotheses have significant impact. </span></strong></p>

Datasets and scripts related to the paper: "Can Generative AI Help us in Open Coding of Software Engineering Data?"

<p>This replication package contains datasets and scripts related to the paper: "<em>Can Generative AI Help us in Open Coding of Software Engineering Data?</em>"</p> <p>The replication package is organized into two directories:</p> <ul> <li> <p><code>manual_analysis</code>: This directory contains all sheets used to perform the manual analysis for RQ1, RQ2, and RQ3.</p> </li> <li> <p><code>stats</code>: This directory contains all datasets, scripts, and results metrics used for the quantitative analyses of RQ1 and RQ2.</p> </li> </ul> <p>In the following, we describe the content of each directory:</p> <h2>manual_analysis</h2> <ul> <li> <p><code>manual_analysis_rq1</code>: This directory contains all sheets used to perform manual analysis for RQ1 (independent and incremental coding).</p> <ul> <li> <p>The sub-directory <code>incremental_coding</code> contains .csv files for all datasets (<code>DL_Faults_COMMIT_incremental.csv</code>, <code>DL_Faults_ISSUE_incremental.csv</code>, <code>DL_Fault_SO_incremental.csv</code>, <code>DRL_Challenges_incremental.csv</code> and <code>Functional_incremental.csv</code>). All these .csv files contain the following columns:</p> <ul> <li><em>Link</em>: The link to the instances</li> <li><em>Prompt</em>: Prompt used as input to GPT-4-Turbo</li> <li><em>ID</em>: Instance ID</li> <li><em>FinalTag</em>: Tag assigned by the human in the original paper</li> <li><em>Chatgpt_output_memory</em>: Output of GPT-4-Turbo with incremental coding</li> <li><em>Chatgpt_output_memory_clean</em>: (only for the DL Faults datasets) output of GPT-4-Turbo considering only the label assigned, excluding the text</li> <li><em>Author1</em>: Label assigned by the first author</li> <li><em>Author2</em>: Label assigned by the second author</li> <li><em>FinalOutput</em>: Label assigned after the resolution of the conflicts</li> </ul> </li> <li> <p>The sub-directory <code>independent_coding</code> contains .csv files for all datasets (<code>DL_Faults_COMMIT_independent.csv</code>, <code>DL_Faults_ISSUE_ independent.csv</code>, <code>DL_Fault_SO_ independent.csv</code>, <code>DRL_Challenges_ independent.csv</code> and <code>Functional_ independent.csv</code>), containing the following columns:</p> <ul> <li><em>Link</em>: The link to the instances</li> <li><em>Prompt</em>: Prompt used as input to GPT-4-Turbo</li> <li><em>ID</em>: Specific ID for the instance</li> <li><em>FinalTag</em>: Tag assigned by the human in the original paper</li> <li><em>Chatgpt_output</em>: Output of GPT-4-Turbo with independent coding</li> <li><em>Chatgpt_output_clean</em>: (only for DL Faults datasets) output of GPT-4-Turbo considering only the label assigned, excluding the text</li> <li><em>Author1</em>: Label assigned by the first author</li> <li><em>Author2</em>: Label assigned by the second author</li> <li><em>FinalOutput</em>: Label assigned after the resolution of the conflicts.</li> </ul> </li> <li> <p>Also, the sub-directory contains sheets with inconsistencies after resolving conflicts. The directory <code>inconsistency_incremental_coding</code> contains .csv files with the following columns:</p> <ul> <li><em>Dataset</em>: The dataset considered</li> <li><em>Human</em>: The label assigned by the human in the original paper</li> <li><em>Machine</em>: The label assigned by GPT-4-Turbo</li> <li><em>Classification</em>: The final label assigned by the authors after resolving the conflicts. Multiple classifications for a single instance are separated by a comma “,”</li> <li><em>Final</em>: final label assigned after the resolution of the incompatibilities</li> </ul> </li> <li> <p>Similarly, the sub-directory <code>inconsistency_independent_coding</code> contains a .csv file with the same columns as before, but this is for the case of independent coding.</p> </li> </ul> </li> <li> <p><code>manual_analysis_rq2</code>: This directory contains .csv files for all datasets (<code>DL_Faults_redundant_tag.csv</code>, <code>DRL_Challenges_redundant_tag.csv</code>, <code>Functional_redundant_tag.csv</code>) to perform manual analysis for RQ2.</p> <ul> <li> <p>The <code>DL_Faults_redundant_tag.csv</code> file contains the following columns:</p> <ul> <li><em>Tags Redundant</em>: tags identified as redundant by GPT-4-Turbo</li> <li><em>Matched</em>: inspection by the authors to see if the tags are redundant matching or not</li> <li><em>FinalTag</em>: final tag assigned by the authors after the resolution of the conflict</li> </ul> </li> <li> <p>The <code>Functional_redundant_tag.csv</code> file contains the same columns as before</p> </li> <li> <p>The <code>DRL_Challenges_redundant_tag.csv</code> file is organized as follows:</p> <ul> <li><em>Tags Suggested</em>: The final tag suggested by GPT-4-Turbo</li> <li><em>Tags Redundant</em>: tags identified as redundant by GPT-4-Turbo</li> <li><em>Matched</em>: inspection by the authors to see if the tags redundant matching or not with the tags suggested</li> <li><em>FinalTag</em>: final tag assigned by the authors after the resolution of the conflict</li> </ul> </li> <li> <p>The sub-directory <code>code_consolidation_mapping_overview</code> contains .csv files (<code>DL_Faults_rq2_overview.csv</code>, <code>DRL_Challenges_rq2_overview.csv</code>, <code>Functional_rq2_overview.csv</code>) organized as follows:</p> <ul> <li><em>Initial_Tags</em>: list of the unique initial tags assigned by GPT-4-Turbo for each dataset</li> <li><em>Mapped_tags</em>: list of tags mapped by GPT-4-Turbo</li> <li><em>Unmatched_tags</em>: list of unmatched tags by GPT-4-Turbo</li> <li><em>Aggregating_tags</em>: list of consolidated tags</li> <li><em>Final_tags</em>: list of final tags after the consolidation task</li> </ul> </li> </ul> </li> <li> <p><code>prompt_for_each_rq</code>: This directory contains: - (i) the history of prompts used in each dataset (<code>prompts_history.txt</code>) -(ii) all final prompt used for the analysis of each dataset, prompt used for incremental coding, prompt used in rq2 to consolidate redundant codes, prompt used in rq3 to create taxonomy (<code>generic_prompt.txt</code>) -(iii) all .csv files in which there are indicate, for each dataset, the link and the prompt used (<code>prompt_DL_Faults_COMMIT.csv</code>, <code>prompt_DL_Faults_ISSUE.csv</code>, <code>prompt_DL_Faults_SO.csv</code>, <code>prompt_DRL_Challenges.csv</code>). For the Functional Dataset .csv file contains, instead, Question, Answer and Prompt used (<code>prompt_Functional.csv</code>)</p> </li> <li> <p><code>rq3</code>: This directory contains the taxonomies obtained from GPT-4-Turbo for the DL Faults and for the DRL Challenges (<code>taxonomy_DL_Faults.txt</code>,<code>taxonomy_DRL_Challenges.txt</code>)</p> </li> </ul> <h2>stats</h2> <ul> <li> <p><code>RQ1</code>: contains script and datasets used to perform metrics for RQ1. The analysis calculates all possible combinations between Matched, More Abstract, More Specific, and Unmatched.</p> <ul> <li><code>RQ1_Stats.ipynb</code> is a Python Jupyter nooteook to compute the RQ1 metrics. To use it, as explained in the notebook, it is necessary to change the values of variables contained in the first code block.</li> <li><code>independent-prompting</code>: Contains the datasets related to the independent prompting. Each line contains the following fields: <ul> <li><em>Link</em>: Link to the artifact being tagged</li> <li><em>Prompt</em>: Prompt sent to GPT-4-Turbo</li> <li><em>FinalTag</em>: Artifact coding from the replicated study</li> <li><em>chatgpt_output_text</em>: GPT-4-Turbo output</li> <li><em>chatgpt_output</em>: Codes parsed from the GPT-4-Turbo output</li> <li><em>Author1</em>: Annotator 1 evaluation of the coding</li> <li><em>Author2</em>: Annotator 2 evaluation of the coding</li> <li><em>FinalOutput</em>: Consolidated evaluation</li> </ul> </li> <li><code>incremental-prompting</code>: Contains the datasets related to the incremental prompting (same format as independent prompting)</li> <li><code>results</code>: contains files for the RQ1 quantitative results. The files are named <code>RQ1\_<<Dataset>>\_<<Prompt method>>\_<<ExcludingNegative>>\_<<MetricAggregation>>.csv</code>, where <em>Dataset</em> is the dataset name, <em>Prompt method</em> indicates whether results are for independent or incremental prompting, <em>Excluding Negatives</em> (for datasets where this applies) whether results have been obtained by excluding negative instances, and <em>MetricAggregation</em> (where it applies) how metrics have been aggregated (macro or weighted average). The files report columns indicating the <em>Dataset</em>, the <em>Matching type</em>, the <em>Accuracy</em>, <em>Precision</em>, <em>Recall</em>, <em>F1 Score</em>, and <em>Cohen's Kappa</em>.</li> </ul> </li> <li> <p><code>RQ2</code>: contains the script used to perform metrics for RQ2, the datasets it uses, and its output.</p> <ul> <li><code>RQ2_SetStats.ipynb</code> is the Python Jupyter notebook to perform the analyses. The scripts takes as input the following types of files, contained in the directory contains the script used to perform the metrics for RQ2. The script takes in input:</li> <li>RQ1 Data Files (<code>RQ1_DLFaults_Issues.csv</code>, <code>RQ1_DLFaults_Commits.csv</code>, and <code>RQ1_DLFaults_SO.csv</code>, joined in a single .csv <code>RQ1_DLFaults.csv</code>). These are the same files used in RQ1.</li> <li>Mapping Files (<code>RQ2_Mappings_DRL.csv</code>, <code>RQ2_Mappings_Functional.csv</code>, <code>RQ2_Mappings_DLFaults.csv</code>). These contain the mappings between human tags (<em>HumanTags</em>), GPT-4-Turbo tags (<em>Final Tags</em>), with indicated the type of matching (<em>MatchType</em>).</li> <li>Additional codes creating during the consolidation (<code>RQ2_newCodes_DRL.csv</code>, <code>RQ2_newCodes_Functional.csv</code>, <code>RQ2_newCodes_DLFaults.csv</code>), annotated with the matching: <em>new code</em>,<em>old code</em>,<em>human code</em>,<em>match type</em></li> <li>Set files (<code>RQ2_Sets_DRL.csv</code>, <code>RQ2_Sets_Functional.csv</code>, <code>RQ2_Sets_DLFaults.csv</code>). Each file contains the following columns: <ul> <li><em>HumanTags</em>: List of tags from the original dataset</li> <li><em>InitialTags</em>: Set of tags from RQ1,</li> <li><em>ConsolidatedTags</em>: Tags that have been consolidated,</li> <li><em>FinalTags</em>: Final set of tags (results of RQ2, used in RQ3)</li> <li><em>NewTags</em>: New tags created during consolidation</li> </ul> </li> <li><code>RQ2_Set_Metrics.csv</code>: Reports the RQ2 output metrics (Precision, Recall, F1-Score, Jaccard).</li> </ul> </li> </ul>

Supplementary materials for the paper "Using AI-Based Coding Assistants in Practice: State of Affairs, Perceptions, and Ways Forward"

<p>This is supplementary materials for the paper "Using AI-Based Coding Assistants in Practice: State of Affairs, Perceptions, and Ways Forward". Please refer to README.txt for more information.</p>

AI at University: raising the attention

<h1>Youtube video "<a href="https://www.youtube.com/watch?v=jLCf3naieFI">Impatto dell’AI su insegnare ed imparare</a>"</h1> <p>Trying to aknowlege the impatc of the so called "Generative Artificial Intelligence" tools on many human activities, particularly on the work environment of graduated students, a short meeting was orgnized pooling some students and professors in Nuoro forestry degree courses on the 15th of October 2024.<br>The video is produced aligning the audio recording with the slide presentation avaliable also as pdf.<br>Video URL: https://www.youtube.com/watch?v=jLCf3naieFI</p>

Data for study "AI-Driven Facial Emotion Analysis for Customer Emotion Recognition - an explanatory study".

<p>This study aims to determine whether it is possible to use a commercial tool like Morphcast SDK AI to identify images that evoke more positive emotions, and thus contribute to a better perception of the hotel facility by customers, and whether it is possible to identify common elements of these photographs. The dataset was collected using  Morphcast SDK AI with users.</p>

Human vs. Machine: How Software Engineers Perceive and Engage with AI-Assisted Code Reviews Compared to Their Peers

<p>This Shared Document Package contains the supplementary materials package for the study, "Human vs. Machine: How Software Engineers Perceive and Engage with AI-Assisted Code Reviews Compared to Their Peers." The package contains all relevant documentation and materials utilized during the study, except interview transcripts. Documents shared in this packes are to assist the readers understand and verify the research protocol and data collection process. It contains:</p> <ol> <li>README file</li> <li>Questionnaire for interview pre-selection</li> <li>Participants’ submitted code snippets</li> <li>Code reviews submitted by participants</li> <li>ChatGPT 4.0-Generated code reviews</li> <li>Complete interview guide</li> <li>Member checking feedback data</li> </ol>

Fractiformers: A New Paradigm in AI Architecture – Whitepapers, Prototypes, and Industry Call-to-Action

<p>This record introduces Fractiformers, a pioneering architecture that redefines traditional transformer models by incorporating fractal-based attention mechanisms, recursive processing, quantum-inspired superposition, and adaptive context management. Developed to address the limitations of transformer-based models in handling long-range dependencies, context adaptation, and computational efficiency, Fractiformers represent a leap forward in AI’s ability to manage complex, dynamic data at scale.</p> <p> </p> <p>Included in this record are detailed whitepapers that outline the theoretical framework, potential applications, and comparative advantages of Fractiformers over traditional transformers. These whitepapers also serve as a call-to-action for GPU manufacturers and AI developers, encouraging collaboration to build hardware and applications optimized for Fractiformers. A prototype implementation is also provided to demonstrate the architecture’s practical benefits, showcasing how Fractiformers outperform conventional models in resource efficiency, contextual coherence, and adaptability.</p> <p> </p> <p>Key Features:</p> <p> </p> <p>• Whitepapers: Comprehensive analysis of Fractiformer architecture, including comparisons with transformers and a breakdown of its unique recursive, fractal, and adaptive components.</p> <p>• Prototype Code: A demonstration prototype to showcase Fractiformers’ performance advantages in handling complex, long-range contexts.</p> <p>• Industry Call-to-Action: An open letter urging GPU manufacturers and AI developers to collaborate on Fractiformer-compatible hardware, with an invitation to participate in joint research and development initiatives.</p> <p> </p> <p>Applications: This record is valuable for AI researchers, GPU manufacturers, and developers aiming to push the boundaries of AI architecture, particularly in fields that demand high-performance, contextually adaptive models such as conversational AI, autonomous systems, and advanced NLP applications.</p>

opencc-by-4.0Sep 2024View details →

[Supplementary Data] PowerModel-AI: A First On-the-fly Machine-Learning Predictor for AC Power Flow Solutions.

<h1><strong>Abstract</strong></h1> <p>The real-time creation of machine-learning models via active or on-the-fly learning has attracted considerable interest across various scientific and engineering disciplines.  These algorithms enable machines to autonomously build models while remaining operational. Through a series of query strategies, the machine can evaluate whether newly encountered data fall outside the scope of the existing training set. In this study, we introduce <em>PowerModel-AI</em>, an end-to-end machine learning software designed to accurately predict AC power flow solutions. We present detailed justifications for our model design choices and demonstrate that selecting the right input features effectively captures the load flow decoupling inherent in power flow equations. Our approach incorporates on-the-fly learning, where power flow calculations are initiated only when the machine detects a need to improve the dataset in regions where the model's performance is sub-optimal, based on specific criteria. Otherwise, the existing model is used for power flow predictions. This study includes analyses of five Texas A&M synthetic power grid cases, encompassing the 14-, 30-, 37-, 200-, and 500-bus systems.  The training and test datasets were generated using <em>PowerModel.jl</em>, an open-source power flow solver/optimizer developed at Los Alamos National Laboratory, NM, USA.</p> <div> <h1><strong>Overview</strong></h1> <p>This dataset, provided as supplementary material for the above-referenced study, includes a comprehensive collection of images (plots) from the study’s analyses, along with Jupyter notebooks containing Python scripts used for the training, validation, and testing phases of PowerModel-AI. Additionally, it includes all training and external test data used in this work, generated via LANL-based open-source power flow solver, PowerModels.jl.</p> <p>The primary objective of this dataset is to ensure full reproducibility of the study’s analyses and facilitate critical examination by the scientific community, thereby maximizing the overall impact of the work.</p> <h1>Directory Structure</h1> <p>The hierarchy of folders and file organization of the dataset is illustrated in the chart below. The directory contains a README.md file which contains the information provided here and a requirements.txt that contains all libraries necessary to run the python scripts or Jupyter notebooks in this directory.  A brief description of the folders and what they contain are provided below:    </p> </div> <div> <strong>.</strong></div> <div>├── <strong>Models/</strong></div> <div>│   ├── _PM_AI_Models/</div> <div>│   │    ├── Type1/</div> <div>│   │    ├── Type2/</div> <div>│   │    └── Type3/</div> <div>│   │</div> <div>│   ├── _PM_AI_Module/</div> <div>│   │    ├── __init__.py</div> <div>│   │    └── PM_Methods.py</div> <div>│   │</div> <div>│   ├── _PM_JL_Data/</div> <div>│   ├── C1_Model/</div> <div>│   ├── C2_Model/</div> <div>│   ├── C3_Model/</div> <div>│   ├── M1_Model/</div> <div>│   ├── M2_Model/  </div> <div>│   └── M3_Model/</div> <div>│</div> <div>├── <strong>NodeSensitivityAnalysis/</strong></div> <div>│   ├── Sensitivity_Plots/</div> <div>│   ├── Sensitivity_PM_JL_Data/</div> <div>│   ├── get_sensitivity_PMJL_data.py</div> <div>│   └── NodeSensitivityAnalysis.ipynb</div> <div>│</div> <div>├── <strong>PlotsForOnTheFlyAnalysis/</strong></div> <div>│   ├── BaseModel_A/</div> <div>│   ├── BaseModel_B/  </div> <div>│   └── BaseModel_C/</div> <div>│  </div> <div>├── <strong>README.md</strong></div> <div>└── <strong>requirements.txt</strong></div> <p> </p> <h2>Files Description </h2> <h3><strong>1.      </strong><strong>Models/</strong></h3> <p><strong><em>_PM_AI_Models/</em></strong> contains the ML models discussed in the manuscript. Type1, Type2, and Type3 refers to the C and M models with numbers "1", "2," and "3". Each Type folder contains individual subfolders for the synthetic grids discussed.</p> <p><strong> <em>_PM_AI_Module/</em></strong> contains a python script that has all the functions used in model training and analysis. It is imported in the Jupyter notebooks in the C and M subfolders in this directory.</p> <p><strong><em>_PM_JL_Data/</em> </strong>contains the following subfolders:</p> <p>a)  <strong> </strong><em>_</em><strong><em><strong>G</strong>eneratePowerModelData/</em></strong> has a python script (<u>get_PowerModelJLData.py)</u> that is used to parse <u>PowerModels.jl</u> to compute AC power flow solutions for different power demand configurations. It also contains a subfolder, <strong><em>BusData_MATLAB/</em></strong>, that has all the synthetic grids used in the study and in MATLAB format.</p> <p>b)   It also contains other subfolders (not shown in the chart above) that contains AC power flow solutions generated using LANL’s PowerModels.jl, for each synthetic grids and other grid-related data.</p> <p>The folders starting with C and M are the control and candidate models (more details in manuscript). Each folder contains Jupyter notebooks (for each power grid) that has python algorithms used in model training and testing, as well as functions to analyze and plot the prediction performance of the models. It accesses (if already trained) or stores (if newly trained) the models in the <strong><em>_PM_AI_Models/</em></strong> directory. Additionally, they contain 2 subfolders (not shown in the chart above): <strong>AbsoluteErrorPlots/</strong> and <strong>PredictionPlots/</strong> were the results (plots) from the analyses are stored. Some of these results are shown in the manuscript (<em>Figures 4</em>,<em> 5</em>, <em>6</em>, and <em>7</em>).</p> <h3><strong>2.      </strong><strong>NodeSensitivityAnalysis/</strong></h3> <p>This folder contains the tools used for the node dependency analysis in Section 3.1.1 of the manuscript.</p> <p>It contains a python script called get_sensitivity_PMJL_data.py, which has similar operation like the <u>get_PowerModelJLData.py</u> script but only compute changes for one bus at a time (see details in manuscript). There is also a Jupyter notebook called <u>NodeSensitivityAnalysis.ipynb </u>that analyzes the bus node dependencies and produces the plots that are shown in Figure 3 in the manuscript. It contains 2 additional sub-folders:</p> <p>a)      <strong> <em>Sensitivity_PM_JL_Data/</em></strong> where PowerModels.jl generated AC power flow solution data are stored, and</p> <p>b)       <em><strong>Sensitivity_Plots/</strong> </em>where the results from <u>NodeSensitivityAnalysis.ipynb</u> are stored.</p> <h3><strong>3.      </strong><strong>PlotsForOnTheFlyAnalysis/</strong></h3> <p>This directory contains only the results for the discussion in Section 3.2 in the manuscript, which is the on-the-fly implementation of PowerModel-AI. The on-the-fly algorithm will be provided and distributed separately in the PowerModel-AI package, which will be publicly available through LANL’s <a href="https://github.com/lanl-ansi" target="_blank" rel="noopener">The Advanced Network Science Initiative</a> (Github). It contains 3 subfolders with similar names but ends with "A", "B," and "C" which correspond to the designations shown and discussed in <em>Figure 8</em> of the manuscript.</p> <h1>Summary</h1> <h3><strong> </strong><strong>Models/</strong></h3> <p><strong>1. _PM_AI_Models/:  </strong></p> <p>Contains the machine learning models discussed in the manuscript. The Type1, Type2, and Type3 folders refer to C and M models labeled "1", "2," and "3". Each type folder includes subfolders for the corresponding synthetic grids analyzed.</p> <div><strong>2. _PM_AI_Module/: </strong> </div> <div>Contains Python scripts for model training and analysis.  PM_Methods.py is the script that defines functions for model training and analysis, which are imported in the Jupyter notebooks in the C and M subfolders.  </div> <div>   </div> <div><strong>3. _PM_JL_Data/:  </strong></div> <div>-_GeneratePowerModelData/: Contains a Python script (get_PowerModelJLData.py) used to parse PowerModels.jl and compute AC power flow solutions for various power demand configurations. This folder also includes BusData_MATLAB/, which holds the synthetic grid data in MATLAB format.  </div> <div>- Other Subfolders: Contain PowerModels.jl AC power flow solutions for each synthetic grid, as well as related data.</div> <div>   </div> <div><strong>4. C1_Model/, C2_Model/, C3_Model/, M1_Model/, M2_Model/, M3_Model/:  </strong></div> <div>These folders contain the control and candidate models (refer to manuscript details). Each folder includes pre-run Jupyter notebooks for model training and analysis, as well as two subfolders:  </div> <div>   - <em>AbsoluteErrorPlots/</em>: Contains saved analysis results for each grid.</div> <div> <strong>  </strong>- <em>PredictionPlots/</em>: Stores model prediction results.  </div> <div>Some results are shown in <em>Figures 4, 5, 6,</em> and <em>7</em> of the manuscript and can be reproduced using the notebooks.</div> <h3>NodeSensitivityAnalysis/</h3> <div><strong>1. Node Dependency Analysis Tools: </strong> </div> <div>   This folder contains scripts and data used for the node dependency analysis in Section 3.1.1 of the manuscript.</div> <div>   </div> <div><strong>2. Scripts:  </strong></div> <div>   - get_sensitivity_PMJL_data.py: Computes power flow changes for individual buses (refer to the manuscript for details).  </div> <div>   - NodeSensitivityAnalysis.ipynb: Analyzes dependencies and generates plots for Figure 3 in the manuscript.  </div> <div>   </div> <div><strong>3. Subfolders:</strong>  </div> <div>   - Sensitivity_PM_JL_Data/: Stores PowerModels.jl data for sensitivity analysis.  </div> <div>   - Sensitivity_Plots/: Contains the generated results from the Jupyter notebook.</div> <h3>PlotsForOnTheFlyAnalysis/</h3> <div><strong>On-the-Fly Learning Results:  </strong>Contains the results for the on-the-fly learning analysis discussed in Section 3.2 of the manuscript.  The subfolders BaseModel_A/, BaseModel_B/, and BaseModel_C/ correspond to the designations in <em>Figure 8</em> of the manuscript. These subfolders contain plots generated during analysis.</div> <h3>Additional Files</h3> <div>- README.md: This file.</div> <div>- requirements.txt: Contains a list of necessary Python libraries required to run the scripts and Jupyter notebooks.</div> <h3>Notes:</h3> <div>- The PowerModel-AI code is publicly available on GitHub.</div> <div>- The supplementary materials include additional Jupyter notebooks and results for each power grid analysis.  </div> <div>- The generated results and predictions in this repository are consistent with those discussed in the manuscript.</div>

AI-Assisted Restoration of Yangshao Painted Pottery Using LoRA and Stable Diffusion

<p>Images related to the paper *AI-Assisted Restoration of Yangshao Painted Pottery Using LoRA and Stable Diffusion* and the specially trained LoRA model for restoring Yangshao pottery patterns.</p>

ChatGPT and AI Perception Change

Open the record for dataset details and reuse information.

Architecturally Significant Requirements and Software Architecture for AI-Based Systems: A Case Study with Document Classification

Open the record for dataset details and reuse information.

An Alien in the Newsroom: AI Anxiety in European and American Newspapers (dataset)

Open the record for dataset details and reuse information.

ADSCAN: AI-Powered Multi-modal Framework for Enhanced Ad Detection and Web Safety

<p>We published codes and datasets (SITE-D, TEXT-D and part of IMG-D) for paper "ADSCAN: AI-Powered Multi-modal Framework for Enhanced Ad Detection and Web Safety."</p>

opencc-byJul 2021View details →

Perceived fairness and perceived transparency of AI systems according to system's characteristics, personality traits and demographic characteristics

<p>We collected data of 3197 users' fairness perception regarding various configurations of a AI-based system in the recruitment domain, as well as, the demographic and personally characteristics of the participants.</p> <p>The dataset includes the following columns:</p> <p><strong>:System characteristics</strong></p> <p> :Certification</p> <p>Uncertificated  system (U)</p> <p>Certificated system (C)</p> <p>:Input data</p> <p>High quality input data (H)</p> <p>Low quality input data (L)</p> <p>:Output</p> <p>Positive outcome (P)</p> <p>Borderline outcome (B)</p> <p>Negative outcome (N)</p> <p>:Explanation style</p> <p>Control- no explanation (CON)</p> <p>Case-based (CAS)</p> <p>Certification-based (CER)</p> <p>Demographic-based (DEM)</p> <p>Input influence-based (INP)</p> <p>Sensitivity-based (SEN)</p> <p><strong>:Demographic characteristics</strong></p> <p>:Gender</p> <p>Female</p> <p>Male</p> <p>:Age</p> <p>18-34</p> <p>35-50</p> <p>50+</p> <p>:Residence</p> <p>Unites states of America</p> <p>India</p> <p>Other</p> <p>:Education level</p> <p>High school degree or less</p> <p>Bachelor's degree</p> <p>Master's or doctoral degree</p> <p>:Employment status</p> <p>Not employed</p> <p>Employed</p> <p>:Income level</p> <p>Above average</p> <p>Average</p> <p>Below average</p> <p><strong>:Personality characteristics</strong></p> <p>(TIPI questionnaire)</p> <p>Extraverted, enthusiastic</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Critical, quarrelsome</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Dependable, self-disciplined</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Anxious, easily upset</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Open to new experiences, complex</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Reserved, quiet</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Sympathetic, warm</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Disorganized, careless</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Calm, emotionally stable</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p>Conventional, uncreative</p> <p>1-7 (1= disagree strongly up to 7= agree strongly)</p> <p><strong>:Participants responses</strong></p> <p>:Fairness evaluation</p> <p>The participants were requested to report their level of perceived fairness (their view about the fairness of the system - at what level they consider the system as a fair system) on a 6-point Likert scale, from "Extremely fair" (represented as 3) to "Extremely unfair" (represented as -3). The option of "neither fair or unfair" (represented as 0) was excluded from the scale.</p> <p>:Transparency evaluation</p> <p>the participants were requested to report their level of perceived transparency (their understanding why the system produced the specific output - at what level they understand why this output was given) on a 6-point Likert scale, from " Thoroughly understand" (represented as 3) to " Thoroughly don't understand" (represented as -3). The option of "neither understand or don't understand" (represented as 0) was excluded from the scale.</p> <p>:Output Expectation</p> <p>The participants were requested to report their expectation for the specific output based on the input they received according to the system's scale, 5-point Likert scale from "Strongly recommended" (represented as 2) to "Strongly not recommended" (represented as -2).</p> <p> </p>

opennotspecifiedJul 2021View details →

Supplementary data: "Revealing drivers and risks for power grid frequency stability with explainable AI"

<p>This repository contains processed data and result files for the paper <a href="https://arxiv.org/abs/2106.04341">Revealing drivers and risks for power grid frequency stability with explainable AI</a>.  The code for producing the processed data and the results is <a href="https://github.com/johkruse/XAI-for-grid-frequency-stability">available at github</a>.</p> <p><strong>Data</strong></p> <p>For each area, the data folder contains the feature and target data used to train the ML model.</p> <ul> <li><em>raw_input_data.h5</em><strong> </strong>:<strong> </strong>The aggregated external features without additional engineered features.</li> <li><em>input_forecast.h5 </em>and<em> input_actual.h5: </em>The day-ahead available (forecast) and ex-post available (actual) data of external features including the engineered features.</li> <li><em>outputs.h5 </em>:  The grid frequency stability indicators.</li> <li><em>version_2021-07-01</em>: Folder containing the training and test sets used for the results.</li> <li><em>documentation_of_data_download</em><strong>: </strong>Plots and information files concerning the ENTSO-E raw data and its aggregation.</li> </ul> <p><strong>Data sources</strong></p> <p>The data for input features (<em>raw_input_data.h5, input_forecast.h5 </em>and<em> input_actual.h5</em>) is derived from ENTSO-E Transparency Platform data [1]. The target data (<em>outputs.h5</em>) is based on power grid frequency recordings from the German Transmission System Operator TransnetBW [2].</p> <p><strong>Results</strong></p> <p>For each area and each target, the result folder comprises the results of hyper-parameter optimization, model prediction and interpretation via SHAP. The results refer to the full model and the restricted model (containing only day-ahead features).</p> <ul> <li><em>cv_results_gtb_full.csv</em> and <em>cv_results_gtb_day_ahead.csv</em> : Performance results for each combination in the hyper-parameter grid search.</li> <li><em>cv_best_params_gtb_full.csv</em> and <em>cv_best_params_gtb_day_ahead.csv</em> : Hyper-parameters used in the final (optimized) model.</li> <li><em>shap_values_gtb_full.npy</em> and <em>shap_interaction_values_gtb_full.npy </em>: First-order SHAP values and second-order SHAP interaction values for the full model.</li> <li><em>y_pred.h5</em> : Predictions of daily profile predictor, full model and day-ahead model.</li> </ul> <p><strong>Disclaimer</strong></p> <p>The data might be subject to copyright or related rights. Please consult the primary data owner.</p>

opencc-by-4.0Aug 2021View details →

Histopathology images for end-to-end AI, based on TCGA-BRCA

<p>These are histopathological images which are derived from the TCGA-BRCA breast cancer histology dataset at https://portal.gdc.cancer.gov/ (please check this website for the original data license). They can be used for end-to-end artificial intelligence (AI) workflows such as DeepMed (https://github.com/KatherLab/deepmed) which aim to predict high-level features directly from digital images with weakly supervised transfer learning. Here, we use two subsets of these digitized images:</p> <p>1) TCGA-BRCA-A2, these are all images from Walter Reed National Military Medical Center (tissue source site code A2, N=100 images) in the TCGA-BRCA database (tcga-brca-a2-deepmed-tiles.zip)</p> <p>2) TCGA-BRCA-E2, these are all images from Roswell Park Comprehensive Cancer Center (tissue source site code E2, N=90 images) in the TCGA-BRCA database (tcga-brca-e2-deepmed-tiles.zip)</p> <p>see also https://gdc.cancer.gov/resources-tcga-users/tcga-code-tables/tissue-source-site-codes </p> <p>The images were preprocessed according to the Aachen Protocol for Deep Learning Histopathology which is available at https://zenodo.org/record/3694994. Specifically, digital whole slide images (SVS format) of hematoxylin & eosin (H&E) stained slides were tessellated (without manual annotations) into tiles of 256x256 px edge length at 1 µm/px. Then, images were color-normalized using the Macenko method as described before (https://www.nature.com/articles/s43018-020-0087-6) and saved as JPEG files. For the A2 cohort, an additional ZIP archive is provided in which only 100 random image tiles are saved for each patient (tcga-brca-a2-deepmed-tiles_100.zip). In addition, we provide a CLINI and a SLIDE table as defined in the "Aachen Protocol". The CLINI table contains clinico-pathological data for all included patients and it is derived from clinical information on www.cbioportal.org as well as from Thorsson et al. (https://pubmed.ncbi.nlm.nih.gov/29628290/). We recommend to use the A2 dataset for training and the E2 dataset for testing. Please cite the relevant papers if you re-use this dataset, more information is available on www.kather.ai</p>

opennotspecifiedSep 2021View details →

Supplementary data: "Secondary control activation analysed and predicted with explainable AI"

<p>This repository contains processed data and result files for the paper  <a href="https://arxiv.org/abs/2109.04802">Secondary control activation analysed and predicted with explainable AI</a> .   The code for producing the processed data and the results is <a href="https://github.com/johkruse/XAI-for-aFRR-activation">available at github</a>.</p> <p><strong>Data</strong></p> <p>The data folder contains the feature and target data used to train the ML model. The data for Germany comprises the following folders and files:</p> <ul> <li><em>raw_input_data.h5</em><strong> </strong>:<strong> </strong>The aggregated external features without additional engineered features.</li> <li><em>inputs_<model_type>.h5 </em><em>:</em> The input features for the different model types used in the paper including the engineered features. Depending on the model type, the input files also contain the IGCC features.</li> <li><em>outputs.h5 </em>:  The activated aFRR volumes in Germany.</li> <li><em>version_2021-08-20</em>: Folder containing the training and test sets used for the results.</li> <li><em>documentation_of_data_download</em>: Information files concerning the ENTSO-E raw data and its aggregation.</li> </ul> <p>In addition to the German time series, the data folder contains the raw input data for the remaining IGCC states. Note that the results contain more model types as actually discussed in the paper.</p> <p><strong>Data sources</strong></p> <p>The data for input features (<em>raw_input_data.h5</em> and <em>input_<model_type>.h5</em>) is derived from ENTSO-E Transparency Platform data [1]. The target data (<em>outputs.h5</em>) is based on publicly available data from the German Transmission System Operators (TSOs) [2].</p> <p><strong>Results</strong></p> <p>The result folder comprises the results of hyper-parameter optimization, model prediction and interpretation via SHAP. The model type, the loss function to train the model and the data set for prediction/interpretation were varied.</p> <ul> <li><em>cv_results_<model_type>_<loss_function>.csv</em> : Performance results for each combination in the hyper-parameter grid search.</li> <li><em>cv_best_params_<model_type>_<loss_function>.csv</em> : Hyper-parameters used in the final (optimized) model.</li> <li><em>shap_values_<data_set>_<model_type>_<loss_function>.npy</em> : First-order SHAP values calculated on different data sets: The train set, the randomized test set and the continuous test set.</li> <li><em>y_pred_<data_set>.h5</em> : Predictions of daily profile predictor and Machine Learning models.</li> </ul> <p><strong>Disclaimer</strong></p> <p>The data might be subject to copyright or related rights. Please consult the primary data owner.</p>

opencc-by-4.0Oct 2021View details →

FOSS4G-IT 2021 - Workshop "QGIS: dal modellatore grafico ai plugin di Processing" (a cura di Federico Gianoli)

<p><strong>Workshop "QGIS: dal modellatore grafico ai plugin di Processing" a cura di Federico Gianoli</strong></p> <p>20 settembre 2021</p> <p>In questo workshop sarà illustrato come configurare il modellatore grafico di QGIS per l'automatizzazione di processi di analisi e l'esportazione del modello come plugin di Processing pronto per essere redistribuito. Per seguire il workshop è sufficiente QGIS LTR.</p> <p> </p> <p>⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=0s">00:00:00</a> | INIZIO PAOLO DABOVE<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=77s">00:01:17</a> | INTRO FEDERICO GIANOLI<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=510s">00:08:30</a> | INTRO PYTHON<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=835s">00:13:55</a> | QGIS e pyQGIS<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=3555s">00:59:15</a> | DOMANDE PRIMA PARTE<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=3955s">01:05:55</a> | MODELLATORE GRAFICO<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=5737s">01:35:37</a> | PLUGIN BUILDER 3<br> ⏰ <a href="https://www.youtube.com/watch?v=2N937a4Rp4w&list=PLk-K8n5iT-AepY_3OOHSjAMFUafBF_Nl3&index=3&t=6798s">01:53:18</a> | DOMANDE FINALI</p> <p><strong>Materiale del Workshop</strong><br> 🔗 <a href="https://github.com/fgianoli/wor_foss4g2021">https://github.com/fgianoli/wor_foss4g2021</a></p> <p><strong>Altri link</strong><br> 🔗 <a href="https://github.com/fgianoli/wor_foss4g2021/blob/main/foss4g.md">https://github.com/fgianoli/wor_foss4g2021/blob/main/foss4g.md</a></p>