用于预测键解离能的图神经网络在数据集间的迁移  

Transferring Graph Neural Network Models for Predicting Bond Dissociation Energy between Datasets

作　　者：霍姚远 江俊 Yao-Yuan Huo;Jun Jiang(Hefei National Laboratory for Physical Sciences at the Microscale,School of Chemistry and Materials Science,University of Science and Technology of China,Hefei 230026,China)

机构地区：[1]中国科学技术大学合肥微尺度物质科学国家研究中心,合肥230026

出　　处：《Chinese Journal of Chemical Physics》2024年第1期51-58,I0118,共9页化学物理学报（英文）

基　　金：supported by the National Key Research and Development Program of China(No.2018YFA0208603);the CAS Project for Young Scientists in Basic Research(No.YSBR-005);the Innovation Program for Quantum Science and Technology(No.2021ZD0303303);the National Natural Science Foundation of China(No.22025304,No.22033007)。

摘　　要：基于机器学习的方法如神经网络已经广泛在化学研究中被用于对化学性质的快速估算.生成高精度的机器学习模型需要高质量的数据集.本文从不同的数据集训练了图神经网络,并验证了模型在数据集间的迁移.结果表明跨数据集的模型预测可以给出精度较低,但相关度良好的结果,其中的误差主要来自系统误差.迁移预测所得的值域与训练集的值域高度相关.不同的键型在迁移预测中的误差大小有所区别,其中C-H键一致地体现出最小的迁移误差.Machine learning(ML)approaches like neural networks have been widely used in chemical researches for fast estimating chemical properties.Generating ML models of good precision requires datasets of high quality,which can be difficult to obtain.In this work,we trained graph neural network(GNN)models from different datasets and verified transferring of the models to other datasets.Our result shows that cross-dataset evaluation can give less accurate but still correlative prediction results on different datasets.Errors are mainly due to systematic errors.The value range of prediction result is highly related to the range of training set.The precisions of different bonds show different distributions.C–H bond constantly gets the highest precision in the tested bonds.

关 键 词：机器学习 图神经网络 模型迁移 

分 类 号：O64[理学—物理化学] TP183[理学—化学]