산술구조연구그룹-계산수리과학연구센터 연합 세미나

일시: 2026년 2월 10일(화) 14:00 ~ 19:00

장소: 강원대학교 춘천캠퍼스 자연과학대학 5호관 201호

프로그램:

• 14:00 ~ 14:30 특강1

•         제목: Birational Equivalence and Compactifications

  • 발표자: 김재현 (이화여자대학교)

  • 초록: Classical objects in algebraic geometry are spaces defined as the common zero loci of polynomial equations. Such spaces, known as varieties, have been studied from various perspectives. In this talk, we focus on birational equivalence, which provides a framework for comparing varieties that are locally but not globally isomorphic. We also introduce the cylinder problem and briefly survey recent results in this area.

  • 
    

• 14:30 ~ 14:45 질의응답 및 휴식

• 
  

• 14:45 ~ 15:15 특강 2

  • 제목: Learning on the Continuum: Consistency beyond Discretization

  • 발표자: 서해윤 (University of Maryland, College Park)

  • 초록: Many phenomena are naturally continuous, but modern machine learning and neural network pipelines typically consume finite representations such as time stamped samples, sequences at a chosen cadence, and point clouds produced by a particular sampling scheme. This talk presents a continuum-aware viewpoint in which the underlying object, such as a curve, a set, a measure, or a function, is primary, and the dataset is a sampled observation of that object. The goal is to explain why representation dependence arises in common neural architectures, why it can produce brittle behavior when sampling schemes change, and what it would mean for a learned model to be consistent under refinement of the representation. Examples will span irregularly sampled time series, continuous-time trajectory modeling and interaction, and neural networks for geometric data.

• 
  

• 15:15 ~ 15:30 질의응답 및 특강 종료