Description

This tutorial is dedicated to the NOMAD artificial-intelligence (AI) toolkit, the platform for running (jupyter) notebooks to analyse with AI tools the data contained in the NOMAD Archive. 

We will cover, in an interactive, hands-on fashion, the several aspects of the AI-toolkit: the query over the NOMAD Archive via the NOMAD API, the basic notebooks for learning AI methods, and the advanced notebooks, where the workflow of relevant publications, in which AI is applied to materials science, can be interactively reproduced and further explored. Furthermore, we will introduce the local AI-toolkit app that allows to run a local version of the notebooks, e.g., to combine own data with the NOMAD Archive data. 

At the end of the first day, few tutorial notebooks will be suggested to be perused by the participants before the second day starts. In the second day, break-out rooms will be organized, and in each room one of the selected tutorial notebooks will be discussed.

Program

April 6 2022	April 7 2022
Talk Sergei V. Kalinin: Bayesian Optimization, structured Gaussian processes, and hypothesis learning for materials and physical discovery 15:00-15:45 CEST	Q&A Luigi Sbailò (unsupervised learning) Daniel Speckhard (decision trees) Andreas Leitherer (deep learning) Šimon Gabaj (electronic-similarity queries) Thomas Purcell and Lucas Foppa (symbolic inference: SISSO and subgroup discovery) 10:00-12:00
Hands-on tutorial Luca Ghiringhelli: Overview of the artificial-intelligence toolkit Luigi Sbailò: Querying the Archive, performing artificial-intelligence analysis, and introducing the AI-toolkit local app Luca Ghiringhelli: Learning artificial-intelligence tools, reproducing publication, overview of selected tutorials and exercises assignment 15:45-18:00 CEST	Q&A Luigi Sbailò (unsupervised learning) Daniel Speckhard (decision trees) Andreas Leitherer (deep learning) Šimon Gabaj (electronic-similarity queries) Thomas Purcell and Lucas Foppa (symbolic inference: SISSO and subgroup discovery) 16:00-18:00