文献类型：期刊文献

中文题名：有机合成中化学反应的机器学习

英文题名：Machine Learning on the Chemical Reaction of Organic Synthesis

作者：张良顺[1]

机构：[1]华东理工大学材料科学与工程学院,上海市先进聚合物材料重点实验室,上海200237

年份：2021

卷号：34

期号：6

起止页码：562

中文期刊名：功能高分子学报

外文期刊名：Journal of Functional Polymers

收录：CSTPCD;;北大核心:【北大核心2020】；CSCD:【CSCD2021_2022】；

语种：中文

中文关键词：化学反应;机器学习;逆合成;图神经网络

外文关键词：chemical reaction;machine learning;retrosynthesis;graph neural network

摘要：化学反应预测及合成路线设计是有机合成领域极具挑战性的问题之一。机器学习是近年来新兴的研究方法。针对有机小分子的化学合成,本文综述了机器学习方法在有机合成(包括化学反应数据的收集、化学反应的预测和合成路线的设计等)领域的进展。对于高度复杂的树脂分子,本文论述了基于机器学习合成路线亟待解决的问题,如数据的不完备和偏倚、缺乏规范化的表示、少样本的机器学习方法等。
It is a challenging task to predict the outcome of chemical reaction and design the route of synthetic planes in the field of organic synthesis.As a novel strategy,the machine-learning approach has extended to study the task of organic synthesis.Focusing on small organic molecules,this work reviews the progress of organic synthesis by virtue of machine-learning approach,including the dataset collection of chemical reaction,the prediction of chemical reaction and the route of synthetic plans.With respect to the molecules of resin with complicated architecture,this work also comments on the challenging issues for organic synthesis based on machine learning,such as the deficiency and bias of dataset,the ill-defined representation of molecular structures and the machine-learning approach with small dataset.