Track Information

Track: Machine Learning Algorithms for Multi-scale Modeling of Molecular Systems

The Machine Learning Algorithms for Multi-scale Modeling of Molecular Systems track focuses on cutting-edge research in enhancing the accuracy and efficiency of molecular (atomistic and coarse-grained, CG) models by applying innovative ML techniques. We invite contributions to state-of-the-art algorithms for effectively approximating the atomistic or CG interactions, numerical assessments of machine learning approaches in diverse material datasets, and novel methodologies for deriving atomistic and/or CG force and energy fields. The track will facilitate discussions on novel ML architectures. We welcome researchers and industry experts to share their findings and contribute their perspectives on overcoming challenges in the field.

This track welcomes papers utilizing diverse research approaches. Topics of interest include (but are not limited to):

• Development of Novel Machine Learning Architectures for Molecular Simulations
• Efficient Training Techniques for ML-Enhanced Atomistic and/or CG Models
• Automated Selection and Optimization of CG Mapping Schemes
• Integrating Physics-Based Constraints in Machine Learning Algorithms
• Transfer Learning for Molecular Dynamics
• Evaluation Metrics and Benchmarks for ML-driven CG Models
• Unsupervised Learning in Atomistic / CG Force Field Generation
• ML Approaches for Accelerating Coarse-Grained Molecular Dynamics
• Data-Driven Methods for Parameterizing Atomistic / CG Potentials
• Machine Learning in Multiscale Modeling: Bridging Atomistic and Coarse-Grained Representations

Track Chair
Vagelis Harmandaris, The Cyprus Institute,