Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea.
Graduate School of Data Science, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea.
J Phys Chem Lett. 2022 Sep 22;13(37):8628-8634. doi: 10.1021/acs.jpclett.2c02293. Epub 2022 Sep 9.
The use of machine learning (ML) is exploding in materials science as a result of its high predictive performance of material properties. Tremendous trainable parameters are required to build an outperforming predictive model, which makes it impossible to retrace how the model predicts well. However, it is necessary to develop a ML model that can extract human-understandable knowledge while maintaining performance for a universal application to materials science. In this study, we developed a deep learning model that can interpret the correlation between surface electronic density of states (DOSs) of materials and their chemisorption property using the attention mechanism that provides which part of DOS is important to predict adsorption energies. The developed model constructs the well-known d-band center theory without any prior knowledge. This work shows that human-interpretable knowledge can be extracted from complex ML models.
机器学习（ML）在材料科学中的应用正在迅速发展，因为它具有很高的材料性能预测能力。为了构建一个表现出色的预测模型，需要使用大量可训练的参数，这使得人们无法追溯模型是如何进行准确预测的。然而，开发一个能够提取人类可理解的知识同时保持性能的 ML 模型是非常有必要的，这样才能将其广泛应用于材料科学。在这项研究中，我们开发了一种深度学习模型，该模型可以使用注意力机制解释材料的表面电子态密度（DOS）与它们的化学吸附性能之间的相关性，注意力机制可以提供 DOS 的哪一部分对于预测吸附能是重要的。所开发的模型无需任何先验知识即可构建出著名的 d 带中心理论。这项工作表明，可以从复杂的 ML 模型中提取人类可理解的知识。
