− HEALTH Artificial intelligence (AI) is rapidly entering dementia clinical practice, offering opportunities across th…+ HEALTH Endocervical gastric-type adenocarcinoma (GAS) is one of the most aggressive subtypes of cervical cancer and…− LIFESTYLE Aquaculture expansion has exacerbated microplastics (MPs) contamination in aquaculture water bodies. MPs read…+ CLIMATE The concentration of iron ions is a crucial indicator for assessing water quality. In this study, nitrogen-do…+ SCIENCE Machine learning interatomic potentials have become an effective method for exploring complex potential energ…± HEALTH Background Artificial intelligence (AI) methods are increasingly used to strengthen policy evaluation in mana…± HEALTH Background Artificial intelligence (AI) applications in managed care pharmacy depend on semantically consiste…± HEALTH Prior authorization (PA) imposes substantial administrative burdens on clinicians, contributing to burnout, d…
TruaceTracing the truth around AIWednesday, September 2, 2026
Subscribe
Science·G Space·Evidence-backed gain·Published 2026-09-02

Data-efficient exploration of atomic clusters via uncertainty quantification on complex potential energy surfaces

Abstract: Machine learning interatomic potentials have become an effective method for exploring complex potential energy surfaces; however, their application to atomic clusters is frequently hindered by the high cost of sampling diverse isomer spaces and the difficulty in ensuring model generalizability across complex energy landscapes. While uncertainty quantification (UQ) offers a pathway to mitigate data scarcity, its efficacy in capturing continuous potential energy surface features and guiding active learning within…

TRV-2026-0959Peer-reviewedPermanent record — cite & verify
Data-efficient exploration of atomic clusters via uncertainty quantification on complex potential energy surfaces

Efficient orchestration of data centers via comprehensive and application-aware trade-off exploration by Bairley, Alan M.. Public domain

The quick read

Machine learning interatomic potentials have become an effective method for exploring complex potential energy surfaces; however, their application to atomic clusters is frequently hindered by the high cost of sampling diverse isomer spaces and the difficulty in ensuring model generalizability across complex energy landscapes. While uncertainty quantification (UQ) offers a pathway to mitigate data scarcity, its efficacy in capturing continuous potential energy surface features and guiding active learning within the complex landscape of clusters remains systematically unverified.

In this study, we developed and evaluated three different UQ frameworks based on advanced UQ methods and integrated with graph neural networks: Bayesian Neural Networks, Evidence Neural Networks (ENN), and Monte Carlo Dropout (MCD).

Main points
  • Machine learning interatomic potentials have become an effective method for exploring complex potential energy surfaces; however, their application to atomic clusters is frequently hindered by the high cost of sampling diverse isomer spaces and the difficulty in ensuring model generalizability across complex energy landscapes.
  • While uncertainty quantification (UQ) offers a pathway to mitigate data scarcity, its efficacy in capturing continuous potential energy surface features and guiding active learning within the complex landscape of clusters remains systematically unverified.
  • In this study, we developed and evaluated three different UQ frameworks based on advanced UQ methods and integrated with graph neural networks: Bayesian Neural Networks, Evidence Neural Networks (ENN), and Monte Carlo Dropout (MCD).
Gain

Data-efficient exploration of atomic clusters via uncertainty quantification on complex potential energy surfaces: Furthermore, active learning driven by MCD significantly reduces the computational overhead of first-principles calculations while maintaining high predictive accuracy.

The rundown

In this study, we developed and evaluated three different UQ frameworks based on advanced UQ methods and integrated with graph neural networks: Bayesian Neural Networks, Evidence Neural Networks (ENN), and Monte Carlo Dropout (MCD). We first validated these models on the MD17 dataset to establish baseline performance, followed by a rigorous assessment on complex cluster systems (Ta2N3- and LaSi24) to probe their decision-making mechanisms in high-dimensional spaces.

Sources

Reader signal

How should this claim be treated?

The debate