차세대가상증강현실연구소에서 아래와 같이 연사를 초청하여 세미나를 진행하오니, 많은 관심과 참여 바랍니다.
- 아 래 -
1. 주제: Automatic Differentiation and Differentiable Physics
2. 일시: 2024.08.01(목) 16:30~17:30
3. 장소: 우정정보관 201호
4. 연사: 엄기원 교수(Telecom Paris, Institut Polytechnique de Paris, France)
Title: Automatic Differentiation and Differentiable Physics
Abstract: Numerical simulation for partial differential equations (PDEs) is a central tool to solve many scientific and engineering problems. To achieve accurate solutions to the given problem, numerical simulation typically involves computation-intensive procedures. Consequently, developing both accurate and efficient numerical methods has been a long-lasting challenge. Recently, machine learning techniques have demonstrated their great capacity for various PDE problems in improving conventional numerical solvers. In this talk, the speaker will first introduce a fundamental element of machine learning algorithms, namely automatic differentiation, and discuss a novel machine learning approach that adopts a differentiable physics framework, which utilizes the automatic differentiation functionality. This framework allows trainable models to interact with PDE solvers in their learning process such that the models can learn better, particularly for temporal evolution learning tasks. The experiments will be presented demonstrating that the proposed approach can address the limitations of conventional PDE solvers. Aiming to reduce numerical errors of given iterative PDE solvers, different learning approaches will be discussed and compared.
Abstract: Numerical simulation for partial differential equations (PDEs) is a central tool to solve many scientific and engineering problems. To achieve accurate solutions to the given problem, numerical simulation typically involves computation-intensive procedures. Consequently, developing both accurate and efficient numerical methods has been a long-lasting challenge. Recently, machine learning techniques have demonstrated their great capacity for various PDE problems in improving conventional numerical solvers. In this talk, the speaker will first introduce a fundamental element of machine learning algorithms, namely automatic differentiation, and discuss a novel machine learning approach that adopts a differentiable physics framework, which utilizes the automatic differentiation functionality. This framework allows trainable models to interact with PDE solvers in their learning process such that the models can learn better, particularly for temporal evolution learning tasks. The experiments will be presented demonstrating that the proposed approach can address the limitations of conventional PDE solvers. Aiming to reduce numerical errors of given iterative PDE solvers, different learning approaches will be discussed and compared.
Speaker: Kiwon Um (Telecom Paris, Institut Polytechnique de Paris, France)
Kiwon Um is an assistant professor in the Computer Graphics group at Telecom Paris, France. Before joining Telecom Paris in October 2019, he worked as a postdoctoral researcher at the Technical University of Munich and Korea University. He earned his PhD degree in computer science and engineering from Korea University. His research interests include physics-based simulations and data-driven approaches with deep learning as well as perceptual evaluation of simulations. He has been investigating more effective ways to simulate natural phenomena via the utilization of refined data and to acquire reliable data for machine learning. Moreover, he has been studying human visual perception for not only computer graphics but also engineering applications while aiming for new evaluation approaches and a better understanding of a variety of numerical methods.