The integration of Machine Learning (ML) techniques into chemistry and materials science has brought about transformative advancements, boosting the pace and precision of predictions. Notable breakthroughs encompass property predictions for materials, the reverse and forward engineering of tailor-made materials, the creation of transferable interatomic potentials, the formulation of differentiable physics models, intelligent exploration of chemical space, and the rationalization of experimental observations, to name just a few success stories. Proposed workshop serves as a platform to unite ML practitioners with diverse backgrounds, spanning both theoretical and experimental domains, in order to foster discussions and forge collaborative efforts among experts from adjacent fields.
The primary sections of the workshop are structured as follows:
● ML Methods for Exploring Non-Equilibrium Processes: This section delves into the applications of ML in understanding non-equilibrium processes such as chemical reactions, materials in extreme conditions, and phase transitions. It offers insights into how ML can be harnessed to enhance our understanding of dynamic events.
● Best Practices in Data Collection: Here, we explore the art of intelligent data collection, focusing on smart sampling, active learning, bias and meta-dynamics, and quantifying uncertainty. These strategies are crucial for ensuring the reliability and accuracy of ML models.
● Meta-Learning Models and Heterogeneous Data Sources: This section navigates the complex landscape of incorporating ML in the face of diverse data sources, including both theoretical and experimental data. It addresses the challenges of harmonizing different data types and sources.
● Development of Differentiable Physics Models: Highlighting the intersection of ML and parametric domain knowledge, this section illuminates the optimization of physics models through machine learning.
● ML-Assisted Design: This section emphasizes the increasing role of ML in assisting design processes, encompassing topics like Quantitative Structure-Activity
Registration fees:
- General Registration - talk, poster or attendance: $350
- Los Alamos National Laboratory (LANL) Employee: $200
- Student or Postdoc (non-LANL): $200