Current browse context:
cs.LG
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: Amplitude Mean of Functional Data on $\mathbb{S}^2$
(Submitted on 29 Jul 2021 (v1), last revised 25 May 2022 (this version, v4))
Abstract: Manifold-valued functional data analysis (FDA) recently becomes an active area of research motivated by the raising availability of trajectories or longitudinal data observed on non-linear manifolds. The challenges of analyzing such data come from many aspects, including infinite dimensionality and nonlinearity, as well as time-domain or phase variability. In this paper, we study the amplitude part of manifold-valued functions on $\mathbb{S}^2$, which is invariant to random time warping or re-parameterization. Utilizing the nice geometry of $\mathbb{S}^2$, we develop a set of efficient and accurate tools for temporal alignment of functions, geodesic computing, and sample mean calculation. At the heart of these tools, they rely on gradient descent algorithms with carefully derived gradients. We show the advantages of these newly developed tools over its competitors with extensive simulations and real data and demonstrate the importance of considering the amplitude part of functions instead of mixing it with phase variability in manifold-valued FDA.
Submission history
From: Bayan Saparbayeva [view email][v1] Thu, 29 Jul 2021 03:11:26 GMT (12175kb,D)
[v2] Fri, 30 Jul 2021 18:11:05 GMT (12175kb,D)
[v3] Wed, 18 Aug 2021 03:10:13 GMT (12175kb,D)
[v4] Wed, 25 May 2022 22:58:32 GMT (19437kb,D)
Link back to: arXiv, form interface, contact.