Jie Jianshu, Hu Zongxiang, Qian Guoyu, Weng Mouyi, Li Shunning, Li Shucheng, Hu Mingyu, Chen Dong, Xiao Weiji, Zheng Jiaxin, Wang Lin-Wang, Pan Feng
School of Advanced Materials, Peking University, Shenzhen Graduate School, Shenzhen 518055, China.
Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley 94720, USA.
Sci Bull (Beijing). 2019 May 15;64(9):612-616. doi: 10.1016/j.scib.2019.04.015. Epub 2019 Apr 5.
Recently, machine learning (ML) has become a widely used technique in materials science study. Most work focuses on predicting the rule and overall trend by building a machine learning model. However, new insights are often learnt from exceptions against the overall trend. In this work, we demonstrate that how unusual structures are discovered from exceptions when machine learning is used to get the relationship between atomic and electronic structures based on big data from high-throughput calculation database. For example, after training an ML model for the relationship between atomic and electronic structures of crystals, we find AgOF, an unusual structure with both Ag and O, from structures whose band gap deviates much from the prediction made by our model. A further investigation on this structure might shed light into the research on anionic redox in transition metal oxides of Li-ion batteries.
最近，机器学习（ML）已成为材料科学研究中广泛使用的技术。大多数工作集中于通过构建机器学习模型来预测规律和总体趋势。然而，新的见解往往是从与总体趋势相悖的例外情况中获得的。在这项工作中，我们展示了在利用机器学习从高通量计算数据库的大数据中获取原子结构与电子结构之间的关系时，如何从例外情况中发现异常结构。例如，在训练了一个关于晶体原子结构与电子结构关系的机器学习模型后，我们从带隙与我们模型预测值偏差很大的结构中发现了AgOF，这是一种同时含有Ag和O的异常结构。对这种结构的进一步研究可能会为锂离子电池过渡金属氧化物中阴离子氧化还原的研究提供线索。
