Ji Hoon Han – Bridging PDEs and machine learning

B232 Seminar Room, IBS 55 Expo-ro Yuseong-gu, Daejeon, Daejeon, Korea, Republic of

Abstract: This talk consists of two main parts. In the first part, I will discuss a numerical method for solving PDEs based on a stochastic representation of the solution. This approach captures the underlying particle dynamics associated with the physical processes described by the PDE. By aggregating information from the particles’ collective exploration, the method iteratively reinforces the approximation toward the solution. I will cover its analysis regarding the trainability and highlight its effectiveness across a broad class of problems, including elliptic equations with interfaces, multiscale structures, and perforated domains, as well as hyperbolic-type problems such as the Eikonal and Burgers equations. In the second part, I will present a method for learning in-between imagery dynamics. This approach integrates PDE models within latent spaces to enhance both learning capability and interpretability. Notably, this method demonstrates robustness in capturing intricate dynamics, such as rotation and outflow, which pose significant

Jae-Kwang Kim – Weight calibration for causal inference and transfer learning

B232 Seminar Room, IBS 55 Expo-ro Yuseong-gu, Daejeon, Daejeon, Korea, Republic of

Abstract: Weight calibration is a popular technique in handling covariate-shift problem in causal inference. It can be viewed as a dual optimization problem for incorporating the implicit regression model. We introduce the generalized entropy calibration as a general tool for weight calibration. Several interesting applications will be introduced in the context of causal inference. Furthermore, weight calibration can be used to transfer learning, which combines information from two different samples, one for source data and the other for target data.

IBS 의생명수학그룹 Biomedical Mathematics Group
기초과학연구원 수리및계산과학연구단 의생명수학그룹
대전 유성구 엑스포로 55 (우) 34126
IBS Biomedical Mathematics Group (BIMAG)
Institute for Basic Science (IBS)
55 Expo-ro Yuseong-gu Daejeon 34126 South Korea
Copyright © IBS 2021. All rights reserved.