Learning the Dynamics of Objects by Optimal Functional Interpolation
- Authors
- Ahn, Jong-Hoon; Kim, In Young
- Issue Date
- Sep-2012
- Publisher
- MIT PRESS
- Citation
- NEURAL COMPUTATION, v.24, no.9, pp.2457 - 2472
- Indexed
- SCIE
SCOPUS
- Journal Title
- NEURAL COMPUTATION
- Volume
- 24
- Number
- 9
- Start Page
- 2457
- End Page
- 2472
- URI
- https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/164788
- DOI
- 10.1162/NECO_a_00325
- ISSN
- 0899-7667
- Abstract
- Many areas of science and engineering rely on functional data and their numerical analysis. The need to analyze time-varying functional data raises the general problem of interpolation, that is, how to learn a smooth time evolution from a finite number of observations. Here, we introduce optimal functional interpolation (OFI), a numerical algorithm that interpolates functional data over time. Unlike the usual interpolation or learning algorithms, the OFI algorithm obeys the continuity equation, which describes the transport of some types of conserved quantities, and its implementation shows smooth, continuous flows of quantities. Without the need to take into account equations of motion such as the Navier-Stokes equation or the diffusion equation, OFI is capable of learning the dynamics of objects such as those represented by mass, image intensity, particle concentration, heat, spectral density, and probability density.
- Files in This Item
-
Go to Link
- Appears in
Collections - 서울 의과대학 > 서울 의공학교실 > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.