How to Train a Shallow Ensemble

Key Points

• This research aims to establish practical guidelines for uncertainty quantification in atomistic machine learning models.
• Developed a model tetrapeptide (Ac-Ala3-NHMe) for testing
• Utilized shallow ensemble methods for uncertainty estimation
• Conducted experiments to assess the reliability of quantification techniques
• Demonstrated improved reliability in uncertainty quantification
• Proved effectiveness of shallow ensemble models in atomistic settings
• Provided actionable guidelines for practitioners in machine learning