Classification and structural characteristics of amorphous materials based on interpretable deep learning

doi:10.1088/1674-1056/acd7d0

中国物理B ›› 2023, Vol. 32 ›› Issue (9): 96101-096101.doi: 10.1088/1674-1056/acd7d0

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Jiamei Cui(崔佳梅)¹, Yunjie Li(李韵洁)¹, Cai Zhao(赵偲)¹, and Wen Zheng(郑文)^1,2,†

1 Institute of Public Safety and Big Data, College of Data Science, Taiyuan University of Technology, Jinzhong 030600, China;
2 Shanxi Engineering Research Center for Intelligent Data Assisted Treatment, Changzhi Medical College, Changzhi 046000, China

收稿日期:2023-03-03 修回日期:2023-05-17 接受日期:2023-05-23 发布日期:2023-09-01
通讯作者: Wen Zheng E-mail:zhengwen@tyut.edu.cn
基金资助:
Project supported by National Natural Science Foundation of China (Grant No. 11702289), the Key Core Technology and Generic Technology Research and Development Project of Shanxi Province, China (Grant No. 2020XXX013), and the National Key Research and Development Project of China.

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Jiamei Cui(崔佳梅)¹, Yunjie Li(李韵洁)¹, Cai Zhao(赵偲)¹, and Wen Zheng(郑文)^1,2,†

1 Institute of Public Safety and Big Data, College of Data Science, Taiyuan University of Technology, Jinzhong 030600, China;
2 Shanxi Engineering Research Center for Intelligent Data Assisted Treatment, Changzhi Medical College, Changzhi 046000, China

Received:2023-03-03 Revised:2023-05-17 Accepted:2023-05-23 Published:2023-09-01
Contact: Wen Zheng E-mail:zhengwen@tyut.edu.cn
Supported by:
Project supported by National Natural Science Foundation of China (Grant No. 11702289), the Key Core Technology and Generic Technology Research and Development Project of Shanxi Province, China (Grant No. 2020XXX013), and the National Key Research and Development Project of China.

摘要/Abstract

摘要： Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder. In this study, we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties. The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the two-dimensional (2D) liquids and amorphous solids of molecular dynamics simulation. The classification process does not make a priori assumptions on the amorphous particle environment, and the accuracy is 92.75%, which is better than other convolutional neural networks. Moreover, our model utilizes the gradient-weighted activation-like mapping method, which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps. We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter. Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems. These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.

关键词: amorphous, interpretable deep learning, image classification, finite scale analysis

Abstract: Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder. In this study, we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties. The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the two-dimensional (2D) liquids and amorphous solids of molecular dynamics simulation. The classification process does not make a priori assumptions on the amorphous particle environment, and the accuracy is 92.75%, which is better than other convolutional neural networks. Moreover, our model utilizes the gradient-weighted activation-like mapping method, which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps. We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter. Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems. These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.

Key words: amorphous, interpretable deep learning, image classification, finite scale analysis

中图分类号: (Disordered solids)

61.43.-j

64.60.aq (Networks) 47.57.-s (Complex fluids and colloidal systems) 64.60.at (Convolution)

Jiamei Cui(崔佳梅), Yunjie Li(李韵洁), Cai Zhao(赵偲), and Wen Zheng(郑文). Classification and structural characteristics of amorphous materials based on interpretable deep learning[J]. 中国物理B, 2023, 32(9): 96101-096101.

参考文献

[1] Stachurski Z H 2011 Materials 4 1564
[2] Wang W H, Dong C and Shek C 2004 Mater. Sci. Eng. 44 45
[3] Inoue A 2000 Acta materialia 48 279
[4] Zhang S, Li Z, Luo K, et al. 2022 Natl. Sci. Rev. 9 1
[5] McHenry M E, Willard M A and Laughlin D E 1999 Prog. Mater. Sci. 44 291
[6] Frumar M, Wagner T, Shimakawa K, et al. 2015 NATO Science for Peace and Security Series C: Environmental Security 139 151
[7] Guinier A 1964 Physics Today 17 70
[8] Tuckerman M 2010 Statistical mechanics: theory and molecular simulation (New York: Oxford university Press)
[9] Steinhardt P J 1983 Phys. Rev. B 28 784
[10] Silver D, Huang A, Maddison C J, et al. 2016 Nature 529 484
[11] Kim J and Park C 2017 Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops pp. 30-38
[12] Tang T T, Zawaski J A, Francis K N, Qutub A A and Gaber M W 2019 Sci. Rep. 9 1
[13] Zhu Q, Samanta A, Li B, Rudd R E and Frolov T 2018 Nat. Commun. 9 1
[14] Doshi-Velez F and Kim B 2017 {arXiv preprint} arXiv: 1702.08608
[15] Murdoch W J, Singh C, Kumbier K, Abbasi-Asl R and Yu B 2019 {arXiv preprint} arXiv: 1901.04592
[16] Chen C, Li O, Barnett A, et al. 2018 Deep Learning for Interpretable Image Recognition arXiv.1806.10574
[17] Selvaraju R R, Cogswell M, Das A, Vedantam R, Parikh D and Batra D 2017 Proceedings of the IEEE International Conference on Computer Vision pp. 618-626
[18] Dietz C, Kretz T and Thoma M 2017 Phys. Rev. E 96 011301
[19] Xiong J, Shi S Q and Zhang T Y 2020 Mater. Design 187 108378
[20] Anker A S, Kjær E T, Juelsholt M, et al. 2022 npj Comput. Mater. 8 213
[21] Du T, Liu H, Tang L, Sorensen S S, Bauchy M and Smedskjaer M M 2021 ACS Nano 15 17705
[22] Mahmoudabadbozchelou M, Kamani K M, Rogers S A and Jamali S 2022 Proc. Natl. Acad. Sci. 119 00278424
[23] Geiger P and Dellago C 2013 J. Chem. Phys. 139 164105
[24] Ziletti A, Kumar D, Scheffler M and Ghiringhelli L M 2018 Nat. Commun. 9 1
[25] Mandal R, Casert C and Sollich P 2022 Nat. Commun. 13 4424
[26] Swanson K, Trivedi S, Lequieu J, Swanson K and Kondor R 2020 Soft Matter 16 435
[27] Zhang K, Li X, Jin Y and Jiang Y 2022 Soft Matter 18 6270
[28] Zhang Y, Zhou W, Ma G, Cheng R and Chang X 2022 Extreme Mechanics Letters 54 101759
[29] Liu H, Tong H and Xu N 2014 Chin. Phys. B 23 116105
[30] Zheng W, Zhang S and Xu N 2018 Chin. Phys. B 27 066102
[31] Liu J, Tong H, Nie Y and Xu N 2020 Chin. Phys. B 29 126302
[32] Hunter G L and Weeks E R 2012 Rep. Prog. Phys. 75 066501
[33] Xu N, O'Hern C S and Kondic L 2005 Phys. Rev. Lett. 94 016001
[34] Tan M and Le Q 2019 Efficientnet: Rethinking Model Scaling for Convolutional Neural Networks pp. 6105-6114
[35] Wang J, Yang L, Huo Z, He W and Luo J 2020 IEEE Access 8 212499
[36] Alhichri H, Alswayed A S, Bazi Y, Ammour N and Alajlan N A 2021 IEEE Access 9 14078
[37] Atila Ü, Uçar M, Akyol K and Uçar E 2021 Ecological Informatics 61 101182
[38] Chetoui M and Akhloufi M A 2020 Explainable Diabetic Retinopathy using EfficientNET
[39] Krizhevsky A, Sutskever I and Hinton G E 2017 Commun. ACM 60 84
[40] Simonyan K and Zisserman A 2014 Computer Science, Computer Vision and Pattern Recognition
[41] He K, Zhang X, Ren S and Sun J 2016 Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Classification and structural characteristics of amorphous materials based on interpretable deep learning

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	Zhao-Ying Xi(奚昭颖), Li-Li Yang(杨莉莉), Lin-Cong Shu(舒林聪), Mao-Lin Zhang(张茂林), Shan Li(李山), Li Shi(史丽), Zeng Liu(刘增), Yu-Feng Guo(郭宇锋), and Wei-Hua Tang(唐为华). The growth and expansive applications of amorphous Ga₂O₃[J]. 中国物理B, 2023, 32(8): 88502-088502.
[2]	Yaocen Wang(汪姚岑), Ziyan Hao(郝梓焱), Yan Zhang(张岩), Xiaoyu Liang(梁晓宇), Xiaojun Bai(白晓军), and Chongde Cao(曹崇德). Formation of L1₀-FeNi hard magnetic material from FeNi-based amorphous alloys[J]. 中国物理B, 2022, 31(4): 46301-046301.
[3]	Hao Sun(孙浩), Junfeng Wang(王俊峰), Lu Tian(田路), Jianjian Gong(巩建建), Zhaojun Mo(莫兆军), Jun Shen(沈俊), and Baogen Shen(沈保根). Magnetic properties and magnetocaloric effect in RE₅₅Co₃₀Al₁₀Si₅ (RE = Er and Tm) amorphous ribbons[J]. 中国物理B, 2022, 31(11): 117503-117503.
[4]	Xiao-Yan Liu(刘晓艳), Lei Wang(王磊), and Yi Tong(童祎). First-principles study of structural and opto-electronic characteristics of ultra-thin amorphous carbon films[J]. 中国物理B, 2022, 31(1): 16102-016102.
[5]	Danhong Li(李丹虹), Changyong Jiang(江昌勇), Hui Li(栗慧), and Mahander Pandey. Crystallization evolution and relaxation behavior of high entropy bulk metallic glasses using microalloying process[J]. 中国物理B, 2021, 30(6): 66401-066401.
[6]	Wen-Si Ai(艾文思), Jie Liu(刘杰), Qian Feng(冯倩), Peng-Fei Zhai(翟鹏飞), Pei-Pei Hu(胡培培), Jian Zeng(曾健), Sheng-Xia Zhang(张胜霞), Zong-Zhen Li(李宗臻), Li Liu(刘丽), Xiao-Yu Yan(闫晓宇), and You-Mei Sun(孙友梅). Degradation of β-Ga₂O₃ Schottky barrier diode under swift heavy ion irradiation[J]. 中国物理B, 2021, 30(5): 56110-056110.
[7]	Wei Zhao(赵伟), Jia-Lin Cheng(成家林), Gong Li(李工), and Xin Wang(王辛). Accurate prediction method for the microstructure of amorphous alloys without non-metallic elements[J]. 中国物理B, 2021, 30(11): 116103-116103.
[8]	Jianing Guo(郭佳宁), Dongli Zhang(张冬利), Mingxiang Wang(王明湘), and Huaisheng Wang(王槐生). Degradation and its fast recovery in a-IGZO thin-film transistors under negative gate bias stress[J]. 中国物理B, 2021, 30(11): 118102-118102.
[9]	陈福川, 代富平, 杨霄熠, 阮莹, 魏炳波. Effect of Sn and Al additions on the microstructure and mechanical properties of amorphous Ti-Cu-Zr-Ni alloys[J]. 中国物理B, 2020, 29(6): 66401-066401.
[10]	唐本镇, 刘晓萍, 李冬梅, 余鹏, 夏雷. Effect of Ni substitution on the formability and magnetic properties of Gd₅₀Co₅₀ amorphous alloy[J]. 中国物理B, 2020, 29(5): 56401-056401.
[11]	周保花, 张福杰, 刘笑, 宋宇, 左旭. Ab initio calculations on oxygen vacancy defects in strained amorphous silica[J]. 中国物理B, 2020, 29(4): 47103-047103.
[12]	张福杰, 周保花, 刘笑, 宋宇, 左旭. Molecular dynamics simulation of atomic hydrogen diffusion in strained amorphous silica[J]. 中国物理B, 2020, 29(2): 27101-027101.
[13]	张彦芳, 陈选虎, 徐阳, 任芳芳, 顾书林, 张荣, 郑有炓, 叶建东. Transition of photoconductive and photovoltaic operation modes in amorphous Ga₂O₃-based solar-blind detectors tuned by oxygen vacancies[J]. 中国物理B, 2019, 28(2): 28501-028501.
[14]	张东, 武辰飞, 徐尉宗, 任芳芳, 周东, 于芃, 张荣, 郑有炓, 陆海. Investigation and active suppression of self-heating induced degradation in amorphous InGaZnO thin film transistors[J]. 中国物理B, 2019, 28(1): 17303-017303.
[15]	汪姚岑, 张岩, Yoshiyuki Kawazoe, 沈军, 曹崇德. Ab initio molecular dynamics simulations of nano-crystallization of Fe-based amorphous alloys with early transition metals[J]. 中国物理B, 2018, 27(11): 116401-116401.