REGISTER HERE https://ml4m.xyz/

PLEASE NOTE THAT REGISTRATION MUST BE DONE THROUGH THE ABOVE WEBSITE.
REGISTRATION THROUGH CECAM WEBSITE WILL NOT BE CONSIDERED VALID.

Data-driven methods have emerged as a novel paradigm to advance materials discovery over the past decade. Machine learning potentials (MLPs) enable the sampling of  trajectories with the same accuracy of high-level electronic structure methods but at a fraction of their cost. [1, 2, 3] MLP have established as a mean to rationalize puzzles previously unapproachable by atomistic simulations. [4, 5, 6] Elsewhere, the chemical and physical properties of large chemical spaces are now screened in a high-throughput fashion by leveraging artificial intelligence methods, materials simulations, and automation protocols. [7, 8]  The screening is not only viable for the case of known structures, but generative models can now autonomously generate previously-unseen, and tailored, molecules and crystals structures with a target property. [9, 10, 11]  Machine learning (ML) methods therefore serve as formidable surrogates to accelerate expensive computational screening, but also to guide experimental screening and extract knowledge from data gathered via high-throughput or from literature. [12, 13, 14] Furthermore, the advances in the theoretical understanding of how machine learning algorithms work is demystifying and surpassing the vision of data-driven approaches as magic black-boxes. [15, 16]

This event aims at building upon the state-of-the-art in the field of machine learning for materials in two ways. The first is to instruct the next generation of young researchers on the latest advancements in methods and applications of AI for material discovery. This scope matches the community-wide need of streamlined and effective programs to train young researchers in data driven methods, which are still too often neglected in Chemistry, Physics, and Material Science curricula. To this end, we will organize tailored introductory lectures, together with an array of hands-on tutorials that focus key aspects of ML for Materials, namely:  Bayesian methods for supervised ML, Neural networks for supervised ML, Unsupervised methods and dimensionality reduction, Materials Descriptors.

Secondly, the workshop will bring forth a discussion on the implications of the latest advancements in data-driven methods on the different sub-areas of Materials discovery. In particular,  two long-standing research questions will be tackled.

1) What are the missing ingredients towards a data-driven or automated route to materials discovery?

To answer this question, we will review a broad array of examples related to ML-accelerated screening using data from DFT calculations, experiments, and the literature. The role of the scientist in developing and evolving screening workflows will be discussed. The key features that make data usable and insightful will be addressed. Renowned experts in the field of Material Science, Catalysis, and Green Energy will present their latest discoveries in the context of data-driven and automated materials discovery and contribute to the discussion with the audience during coffee breaks and panel discussions.

2) Can we reconcile the dichotomy between Physics-aware and Deep ML Methods and get the best of both worlds? 

To answer this question, we will review the latest results obtained by the two approaches, also as a function of their interpretability. Also in this case, leading figures in the field will provide their expertise, recent advancements and prompt discussion among the participants. Nurturing discussion is essential for a multi-disciplinary insight, needed to overcome the complexity inherent to materials discovery. To achieve this, the event will gather emerging leaders and established ones from varied disciplines (Physics, Chemistry, Engineering, Mathematics) to discuss the latest developments specific to their subject of research. The cross-fertilization of ideas and methods is expected to be beneficial to overcome bottlenecks in specific subjects,  highlight common trends, and address unresolved issues across sub-themes.