TY - GEN
T1 - Physics Knowledge Discovery via Neural Differential Equation Embedding
AU - Xue, Yexiang
AU - Nasim, Md
AU - Zhang, Maosen
AU - Fan, Cuncai
AU - Zhang, Xinghang
AU - El-Azab, Anter
PY - 2021
Y1 - 2021
N2 - Despite much interest, physics knowledge discovery from experiment data remains largely a manual trial-and-error process. This paper proposes neural differential equation embedding (NeuraDiff), an end-to-end approach to learn a physics model characterized by a set of partial differential equations directly from experiment data. The key idea is the integration of two neural networks – one recognition net extracting the values of physics model variables from experimental data, and the other neural differential equation net simulating the temporal evolution of the physics model. Learning is completed by matching the outcomes of the two neural networks. We apply NeuraDiff to the real-world application of tracking and learning the physics model of nano-scale defects in crystalline materials under irradiation and high temperature. Experimental results demonstrate that NeuraDiff produces highly accurate tracking results while capturing the correct dynamics of nano-scale defects.
AB - Despite much interest, physics knowledge discovery from experiment data remains largely a manual trial-and-error process. This paper proposes neural differential equation embedding (NeuraDiff), an end-to-end approach to learn a physics model characterized by a set of partial differential equations directly from experiment data. The key idea is the integration of two neural networks – one recognition net extracting the values of physics model variables from experimental data, and the other neural differential equation net simulating the temporal evolution of the physics model. Learning is completed by matching the outcomes of the two neural networks. We apply NeuraDiff to the real-world application of tracking and learning the physics model of nano-scale defects in crystalline materials under irradiation and high temperature. Experimental results demonstrate that NeuraDiff produces highly accurate tracking results while capturing the correct dynamics of nano-scale defects.
KW - Nano-scale materials science
KW - Neural differential equation embedding
KW - Physics knowledge discovery
UR - http://www.scopus.com/inward/record.url?scp=85115698531&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85115698531&origin=recordpage
U2 - 10.1007/978-3-030-86517-7_8
DO - 10.1007/978-3-030-86517-7_8
M3 - 32_Refereed conference paper (with host publication)
SN - 978-3-030-86516-0
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence)
SP - 118
EP - 134
BT - Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track
A2 - Dong, Yuxiao
A2 - Kourtellis, Nicolas
A2 - Hammer, Barbara
A2 - Lozano, Jose A.
PB - Springer
CY - Cham
T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2021)
Y2 - 13 September 2021 through 17 September 2021
ER -