编者按
Data Transfer Approaches to Improve Seq-to-Seq
Retrosynthesis 是2020年10月发表在Computer Science上的关于逆合成路线规划一篇文章。首次研究基于数据驱动的逆合成方法探究,将不基于模板和基于模板的逆合成进行比对,在预训练和微调过程中探究不同数据的增强方式对于逆合成中模型的影响大小,以供读者更好的理解文献。
This article explains the importance of data sets in the field of retrosynthesis. Because in addition to the issues that need to be considered in the model architecture itself, the number of available data sets (samples) and quality (less noisy, damaged samples) usually determine the final performance of the machine learning model. However, there are usually only a few large and high-quality supervised training data sets that are publicly available. Therefore, this paper mainly studies the impact of data transmission on the improved inverse synthesis model, and proposes three standard data transm