Effective practices for analyzing, sharing, and utilizing research data.

Federated Learning

Federated learning is a decentralized machine learning approach where multiple entities collaborate to train a model without sharing their data. Instead of centralizing data, the model is trained locally on each node, and only model updates are shared, ensuring sensitive data remains at its source. This approach enhances data privacy and compliance with regulations, reducing the risk of data breaches. By leveraging data from multiple nodes, federated learning improves model robustness and generalizability, offering insights across diverse scenarios. The tool is available under an open-source license.

Skill Level: Beginner

Risk Assessment

This course offers an overview of the Risk Assessment tool, designed to evaluate risks associated with sharing FCT data. Participants will be guided through assessing impact levels, threat likelihoods, and mitigation actions. In a hands-on training session, users will conduct a sample risk assessment for a hypothetical request involving personal data from traffic cameras. The results will be presented in a color-coded table and an interactive radar chart, with options for sharing or utilizing the findings effectively.

Skill Level: Beginner

Data Quality Assessment

This course centers on the Data Quality Assessment (DQA) tool within the LAGO project, which evaluates data quality across multiple formats, including tabular data, images, videos, and annotated datasets. Participants will learn how the tool generates detailed reports and warnings, enabling them to analyze variables, correlations, missing values, and duplicate rows in datasets. It also facilitates comparisons between similar datasets and assessments of image and video quality. By using this tool, learners can ensure a thorough examination of data integrity, ensuring compliance with project requirements.

Skill Level: Beginner