English

Self-Supervised Learning for Ordered Three-Dimensional Structures

Machine Learning 2024-11-25 v1

Abstract

Recent work has proven that training large language models with self-supervised tasks and fine-tuning these models to complete new tasks in a transfer learning setting is a powerful idea, enabling the creation of models with many parameters, even with little labeled data; however, the number of domains that have harnessed these advancements has been limited. In this work, we formulate a set of geometric tasks suitable for the large-scale study of ordered three-dimensional structures, without requiring any human intervention in data labeling. We build deep rotation- and permutation-equivariant neural networks based on geometric algebra and use them to solve these tasks on both idealized and simulated three-dimensional structures. Quantifying order in complex-structured assemblies remains a long-standing challenge in materials physics; these models can elucidate the behavior of real self-assembling systems in a variety of ways, from distilling insights from learned tasks without further modification to solving new tasks with smaller amounts of labeled data via transfer learning.

Keywords

Cite

@article{arxiv.2411.14680,
  title  = {Self-Supervised Learning for Ordered Three-Dimensional Structures},
  author = {Matthew Spellings and Maya Martirossyan and Julia Dshemuchadse},
  journal= {arXiv preprint arXiv:2411.14680},
  year   = {2024}
}

Comments

Version as submitted to the Learning on Graphs Conference 2022, with small clarifying edits