A benchmark study of machine learning methods for molecular electronic transition: Tree-based ensemble learning versus graph neural network,Bulletin of the Korean Chemical Society

当前位置： X-MOL 学术 › Bull. Korean Chem. Soc. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

A benchmark study of machine learning methods for molecular electronic transition: Tree-based ensemble learning versus graph neural network
Bulletin of the Korean Chemical Society ( IF 2.3 ) Pub Date : 2022-01-10 , DOI: 10.1002/bkcs.12468
Beomchang Kang ₁ , Chaok Seok ₁ , Juyong Lee ₂

Affiliation

Fluorophores play crucial roles in chemical and biological imaging. An efficient computational model that evaluates the electronic properties of molecules accurately would be a useful tool for discovering novel fluorophores. Tree-based ensemble and graph neural network (GNN) methods have been regarded as attractive models for predicting molecular properties. Here, we present a benchmark test using three tree-based ensemble methods (Random Forest, LightGBM, and XGBoost) and three GNNs (directed message passing neural network [D-MPNN], attention message passing neural network [AMPNN], and DimeNet++) for predicting electronic transition properties such as excitation energies and oscillator strengths. From our benchmark, DimeNet++ was identified as the most accurate model to predict electronic transition properties. The average root mean square error of DimeNet++ for predicting the HOMO–LUMO gap was 0.11 eV whereas those of the other methods exceeded 0.3 eV. D-MPNN predicted fastest without sacrificing accuracy. Our results show that DimeNet++ and D-MPNN may serve as helpful evaluators for novel fluorophore design when combined with molecular generation methods.

中文翻译：

分子电子跃迁机器学习方法的基准研究：基于树的集成学习与图神经网络

荧光团在化学和生物成像中起着至关重要的作用。准确评估分子电子特性的有效计算模型将是发现新型荧光团的有用工具。基于树的集成和图神经网络 (GNN) 方法已被视为预测分子特性的有吸引力的模型。在这里，我们使用三种基于树的集成方法（随机森林、LightGBM 和 XGBoost）和三种 GNN（定向消息传递神经网络 [D-MPNN]、注意力消息传递神经网络 [AMPNN] 和 DimeNet++）进行基准测试用于预测电子跃迁特性，例如激发能量和振荡器强度。在我们的基准测试中，DimeNet++ 被确定为预测电子跃迁特性的最准确模型。DimeNet++ 用于预测 HOMO-LUMO 间隙的平均均方根误差为 0.11 eV，而其他方法的均方根误差超过 0.3 eV。D-MPNN 在不牺牲准确性的情况下预测得最快。我们的结果表明，当与分子生成方法相结合时，DimeNet++ 和 D-MPNN 可以作为新型荧光团设计的有用评估器。

更新日期：2022-01-10

点击分享查看原文

点击收藏

阅读更多本刊新发论文本刊介绍/投稿指南