【海韵讲座】2021年第28期- High-order Differentiable Autoencoder for Nonlinear Model Reduction
报告题目：High-order Differentiable Autoencoder for Nonlinear Model Reduction
主讲人：杨垠，美国克莱姆森大学副教授
时间：2021年6月28日（星期一）9:00-10:30
地点：腾讯会议
点击链接入会，或添加至会议列表：https://meeting.tencent.com/s/O8oO0MJjPRZZ
会议 ID：775 577 106
手机一键拨号入会
+8675536550000,,775577106# (中国大陆)
+85230018898,,,2,775577106# (中国香港)
摘要：
In this talk, a new avenue for exploiting deep neural networks to improve physics-based simulation will be introduced. Specifically, we integrate the classic Lagrangian mechanics with a deep autoencoder to accelerate elastic simulation of deformable solids. Due to the inertia effect, the dynamic equilibrium cannot be established without evaluating the second-order derivatives of the deep autoencoder network. This is beyond the capability of off-the-shelf automatic differentiation packages and algorithms, which mainly focus on the gradient evaluation. Solving the nonlinear force equilibrium is even more challenging if the standard Newton's method is to be used. This is because we need to compute a third-order derivative of the network to obtain the variational Hessian. We attack those difficulties by exploiting complex-step finite difference, coupled with reverse automatic differentiation. This strategy allows us to enjoy the convenience and accuracy of complex-step finite difference and in the meantime, to deploy complex-value perturbations as collectively as possible to save excessive network passes.
报告人简介：
Dr. Yin Yang is currently an Associate Professor with School of Computing, Clemson University. Before that, he was a faculty member with the department of Electrical and Computer Engineering, University of New Mexico. He received Ph.D. degree of Computer Science from the University of Texas, Dallas in 2013 (the awardee of David Daniel Fellowship Prize). During his Ph.D., he was also a Research Assistant at UT Southwestern Medical Center. He was a research intern in Microsoft Research Asia in 2012. His research aims to develop customized and efficient computational methods for computer graphics, animation, vision, fabrication, robotics, autonomous driving etc. He received NSF CRII award and NSF CAREER award in 2015 and 2019 respectively.
邀请人：软件工程系 姚俊峰教授