A community-powered search of machine learning strategy space to find NMR property prediction models
This work addresses the difficulty for physical scientists in selecting optimal ML strategies for domain-specific predictions, though it is incremental as it applies existing community competition methods to a new dataset.
The authors tackled the challenge of efficiently exploring the vast machine learning strategy space for predicting NMR properties in molecules by organizing a community-powered Kaggle competition, which within 3 weeks produced models matching their previous best efforts and a meta-ensemble model achieving 7-19x better accuracy than prior state-of-the-art.
The rise of machine learning (ML) has created an explosion in the potential strategies for using data to make scientific predictions. For physical scientists wishing to apply ML strategies to a particular domain, it can be difficult to assess in advance what strategy to adopt within a vast space of possibilities. Here we outline the results of an online community-powered effort to swarm search the space of ML strategies and develop algorithms for predicting atomic-pairwise nuclear magnetic resonance (NMR) properties in molecules. Using an open-source dataset, we worked with Kaggle to design and host a 3-month competition which received 47,800 ML model predictions from 2,700 teams in 84 countries. Within 3 weeks, the Kaggle community produced models with comparable accuracy to our best previously published "in-house" efforts. A meta-ensemble model constructed as a linear combination of the top predictions has a prediction accuracy which exceeds that of any individual model, 7-19x better than our previous state-of-the-art. The results highlight the potential of transformer architectures for predicting quantum mechanical (QM) molecular properties.