Ph.D student at UC Berkeley. Deep Learning methods for 3D Geometries and Physical Systems.
About:
I’m a fourth-year Ph.D student at UC Berkeley advised by Prof. Philip Marcus, with affiliations with the Data Analytics group at NERSC, Lawrence Berkeley National Lab. My research interest is in Machine Learning and Deep Learning techniques, applied to 3D geometry. In particular, I am interested in leveraging advances in 3D learning for applications in a variety of physical and engineering systems, examples include omnidirectional image segmentation, climate pattern detection and aerodynamical shape optimization.
Prior to joining UC Berkeley, I got my Bachelor’s degree from Cornell University, as well as a joint degree from Zhejiang University.
Research Page:
Please find more information in my personal research page.
Research at CFD Lab with Prof. Marcus:
Full-Length Articles
(2020) Enforcing Physical Constraints in CNNs through Differentiable PDE Layer, ICLR 2020 Workshop on Integration of Deep Neural Models and Differential Equations, pdf
(2020) Meshfreeflownet: A physics-constrained deep continuous space-time super-resolution framework, SC20: International Conference for High Performance Computing, Networking, Storage and Analysis, p. 1-15, IEEE, pdf
(2019) DDSL: Deep differentiable simplex layer for learning geometric signals, Proceedings of the IEEE/CVF International Conference on Computer Vision, p. 8769-8778, pdf
(2018) Convolutional Neural Networks on Non-uniform Geometrical Signals Using Euclidean Spectral Transformation, International Conference on Learning Representations, pdf
(2018) Spherical CNNs on Unstructured Grids, International Conference on Learning Representations, pdf
(2018) Finding the optimal shape of the leading-and-trailing car of a high-speed train using design-by-morphing, Computational Mechanics 62(1), p. 23-45, pdf, doi:10.1007/s00466-017-1482-4
(2017) Hierarchical Detail Enhancing Mesh-Based Shape Generation with 3D Generative Adversarial Network, arXiv preprint arXiv:1709.07581, pdf
Conference Papers/Abstracts
(2020) A deep learning based physics informed continuous spatio temporal super-resolution framework, APS Division of Fluid Dynamics Meeting Abstracts, p. S01–033, pdf
(2019) Neural Network Optimization Under Partial Differential Equation Constraints, APS Division of Fluid Dynamics Meeting Abstracts, p. C17–008, url
(2018) Neural Network Powered Adjoint Methods-Gradient Based Shape Optimization with Deep Learning, APS Division of Fluid Dynamics Meeting Abstracts 63, p. F32-002, url
(2018) Bridging simulation and deep learning-convolutional neural networks on unstructured grids, APS Division of Fluid Dynamics Meeting Abstracts 63, p. F32-005, url
(2018) Deep learning on the Sphere: Convolutional Neural Network on Unstructured Mesh, AGU Fall Meeting Abstracts 2018, p. IN33A–01, url
(2017) Drag Reduction of an Airfoil Using Deep Learning, APS Division of Fluid Dynamics Meeting Abstracts, p. D31–008, url
(2016) Shape Optimization of A Turbine-99 Draft Tube Using Design-by-Morphing, APS Division of Fluid Dynamics Meeting Abstracts, p. L4–004, url