This repository is intended to create a series of notebooks that illustrate the FAIR principles in the context of High Energy Physics Data and AI models. The FAIR acronym stands for Findable, Accessible, Interoperable, and Reusable- benchmarking a set of desired qualities that ensure transparent and objective scientific data management.

List of Notebooks:

• 01-Intro2FAIR.ipynb: Introduces the FAIR principles and a set of metrics used to evaluate FAIRness of datasets
• 02-FAIRCheck-MNIST.ipynb: Explores the MNIST dataset and its FAIRness using the metrics introduced in the previous notebook
• 03-FAIRCheck-CDMS.ipynb: Explores the superCDMS dataset and its FAIRness
• 04a-CDMS-LR.ipynb: Explores linear regression along with Ridge and Lasso Regularizations and preservation of these models
• 04b-CDMS-LR-FAIR.ipynb: Explores how the model can be built, preserved, and reused in line with the FAIR principles
• 05a-CDMS-PCA.ipynb: Explores Principal Component Analysis based linear regression. Building model scripts in a FAIR way and demonstrating its use via a notebook is left as an exercise
• 07a-CDMS_NNRegressor.ipynb: Explores a Neural Network based approach to solve the regression from on the CDMS dataset
• 07b-CDMS_NNRegressor-FAIR.ipynb: Explores how the NN based model can be built, preserved, and reused in line with the FAIR principles
• 08a-CDMS_NNVAE.ipynb: An introduction to Variation Auto Encoder and its implementation for the CDMS dataset
• 08b-CDMS_NNVAE.ipynb: Explores the model building and preservation in a FAIR way. Ends with a series of exercise that will (a) allow the user to explore how model metadata can be used to reuse a preserved model and (b) modify current model to incorporate better flexibility for users