Deep graph kernel learning for material & atomic level uncertainty quantification in adsorption energy prediction
Open Access
- 24 March 2026
- journal article
- research article
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 5 (4) , 1568-1578
Abstract
Deep graph kernel learning combines GNN backbones with Gaussian processes to deliver calibrated, atomic-level uncertainty quantification for adsorption energy prediction, outperforming ensemble and evidential methods for catalyst discovery.Keywords
All Related Versions
Funding Information
- Research Corporation for Science Advancement (SA-SM3-2024-059b)
This publication has 31 references indexed in Scilit:
- Fine-Tuning Graph Neural Networks via Active Learning: Unlocking the Potential of Graph Neural Networks Trained on Nonaqueous Systems for Aqueous CO2 ReductionJournal of Chemical Theory and Computation, 2025
- Single-model uncertainty quantification in neural network potentials does not consistently outperform model ensemblesnpj Computational Materials, 2023
- Evidential Deep Learning for Guided Molecular Property Prediction and DiscoveryACS Central Science, 2021
- Open Catalyst 2020 (OC20) Dataset and Community ChallengesACS Catalysis, 2021
- Generalized combination rule for evidential reasoning approach and Dempster–Shafer theory of evidenceInformation Sciences, 2021
- Uncertainty quantification in molecular simulations with dropout neural network potentialsnpj Computational Materials, 2020
- Recent advances and applications of machine learning in solid-state materials sciencenpj Computational Materials, 2019
- Catalysis-Hub.org, an open electronic structure database for surface reactionsScientific Data, 2019
- To address surface reaction network complexity using scaling relations machine learning and DFT calculationsNature Communications, 2017
- Universality in Heterogeneous CatalysisJournal of Catalysis, 2002