BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Biomedical Mathematics Group - ECPv6.15.20//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://www.ibs.re.kr/bimag
X-WR-CALDESC:Events for Biomedical Mathematics Group
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:20220101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Seoul:20230217T140000
DTEND;TZID=Asia/Seoul:20230217T160000
DTSTAMP:20260426T064235
CREATED:20230127T011541Z
LAST-MODIFIED:20230214T144610Z
UID:7278-1676642400-1676649600@www.ibs.re.kr
SUMMARY:Hyeontae Jo\, Characterizing possible failure modes in physics-informed neural networks
DESCRIPTION:We will discuss about “Characterizing possible failure modes in physics-informed neural networks”\, Krishnapriyan\, Aditi\, et al.\, Advances in Neural Information Processing Systems 34 (2021): 26548-26560. \nAbstract \n\n\n\n\n\n\nRecent work in scientific machine learning has developed so-called physics-informed neural network (PINN) models. The typical approach is to incorporate physical domain knowledge as soft constraints on an empirical loss function and use existing machine learning methodologies to train the model. We demonstrate that\, while existing PINN methodologies can learn good models for relatively trivial problems\, they can easily fail to learn relevant physical phenomena for even slightly more complex problems. In particular\, we analyze several distinct situations of widespread physical interest\, including learning differential equations with convection\, reaction\, and diffusion operators. We provide evidence that the soft regularization in PINNs\, which involves PDE-based differential operators\, can introduce a number of subtle problems\, including making the problem more ill-conditioned. Importantly\, we show that these possible failure modes are not due to the lack of expressivity in the NN architecture\, but that the PINN’s setup makes the loss landscape very hard to optimize. We then describe two promising solutions to address these failure modes. The first approach is to use curriculum regularization\, where the PINN’s loss term starts from a simple PDE regularization\, and becomes progressively more complex as the NN gets trained. The second approach is to pose the problem as a sequence-to-sequence learning task\, rather than learning to predict the entire space-time at once. Extensive testing shows that we can achieve up to 1-2 orders of magnitude lower error with these methods as compared to regular PINN training. \n 
URL:https://www.ibs.re.kr/bimag/event/2023-02-17-jc/
LOCATION:B378 Seminar room\, IBS\, 55 Expo-ro Yuseong-gu\, Daejeon\, 34126\, Korea\, Republic of
CATEGORIES:Journal Club
ORGANIZER;CN="Jae Kyoung Kim":MAILTO:jaekkim@kaist.ac.kr
END:VEVENT
END:VCALENDAR