npj Computational Materials

Deep-learning-based inverse design model for intelligent discovery of organic molecules （基于深度学习的逆向设计模型：用于有机分子的智能发现）
Kyungdoc Kim, Seokho Kang, Jiho Yoo, Youngchun Kwon, Youngmin Nam, Dongseon Lee, Inkoo Kim, Youn-Suk Choi, Yongsik Jung, Sangmo Kim, Won-Joon Son, Jhunmo Son, Hyo Sug Lee, Sunghan Kim, Jaikwang Shin & Sungwoo Hwang
npj Computational Materials 4:67 (2018)
doi:s41524-018-0128-1
Published online:03 December 2018
Abstract| Full Text | PDF OPEN

摘要:高性能功能材料的发现对于克服现代工业中的技术问题至关重要。人们不仅在实验上，而且在材料设计的层次对加速和促进该过程已作了广泛的努力。最近，机器学习以其能为材料的有效探索提供理性指导，既能避免耗时的迭代或前人的知识积累，而引起了众多的关注。在这方面，本研究就此开发了基于深度编码器-解码器架构的逆向设计模型，用于目标分子设计。受神经机器语言翻译的启发，深度神经网络编码器能提取分子结构与其材料特性之间的隐藏特征，而递归神经网络解码器能将提取的特征重建为具有目标性质的新分子结构。在材料设计任务中，所提出的完全数据驱动的方法成功地从给定的数据库中学习了设计规则，并通过创建新的配体和组合规则，为磷光有机发光二极管提供了有希望的光吸收分子和主体材料。　　

Abstract:The discovery of high-performance functional materials is crucial for overcoming technical issues in modern industries. Extensive efforts have been devoted toward accelerating and facilitating this process, not only experimentally but also from the viewpoint of materials design. Recently, machine learning has attracted considerable attention, as it can provide rational guidelines for efficient material exploration without time-consuming iterations or prior human knowledge. In this regard, here we develop an inverse design model based on a deep encoder-decoder architecture for targeted molecular design. Inspired by neural machine language translation, the deep neural network encoder extracts hidden features between molecular structures and their material properties, while the recurrent neural network decoder reconstructs the extracted features into new molecular structures having the target properties. In material design tasks, the proposed fully data-driven methodology successfully learned design rules from the given databases and generated promising light-absorbing molecules and host materials for a phosphorescent organic light-emitting diode by creating new ligands and combinatorial rules.

Editorial Summary

Organic optoelectronics: Efficient molecules designed by your computer （有机光电子学：由你计算机设计出的高效分子）

告诉你的计算机你所需要的材料属性，它就会设计出你正需寻找的分子。来自韩国三星和成均馆大学的Kyungdoc Kim及其同事，开发了两种计算机算法，可以为此计算设计目的而共同工作。第一种算法能查看已知有机分子及其性能的数据库，并找到描述结构/性能关系的抽象规则;第二种算法使用这些规则来设计预期的具有相同目标性质的新分子结构。用这种方法，他们已经提出了能够吸收所需波长的分子，以及用于实现高效、稳定的蓝光有机显示器材料。该技术可用于发现应用范围更广的相关全新分子和设计规则。

Tell your computer the materials properties you need, and it will design the molecule you are looking for. Kyungdoc Kim and colleagues from Samsung and Sungkyunkwan University, Republic of Korea, have developed two computer algorithms that work together for this purpose. The first algorithm looks at a database of known organic molecules and their properties, and finds abstract rules to describe the structure/property relationships; the second one uses these rules to design new molecular structures expected to have the same targeted properties. Using this approach, the researchers have already proposed molecules able to absorb light of a desired color, and materials for the realization of stable and efficient organic displays emitting in the blue. The technique might be applied to discover novel molecules and design rules relevant to a broader range of applications.