Fast matrix computations for functional additive models
From MaRDI portal
Publication:5963540
DOI10.1007/S11222-014-9490-0zbMATH Open1331.62027arXiv1402.4984OpenAlexW1992731132MaRDI QIDQ5963540FDOQ5963540
Publication date: 22 February 2016
Published in: Statistics and Computing (Search for Journal in Brave)
Abstract: It is common in functional data analysis to look at a set of related functions: a set of learning curves, a set of brain signals, a set of spatial maps, etc. One way to express relatedness is through an additive model, whereby each individual function is assumed to be a variation around some shared mean . Gaussian processes provide an elegant way of constructing such additive models, but suffer from computational difficulties arising from the matrix operations that need to be performed. Recently Heersink & Furrer have shown that functional additive model give rise to covariance matrices that have a specific form they called quasi-Kronecker (QK), whose inverses are relatively tractable. We show that under additional assumptions the two-level additive model leads to a class of matrices we call restricted quasi-Kronecker, which enjoy many interesting properties. In particular, we formulate matrix factorisations whose complexity scales only linearly in the number of functions in latent field, an enormous improvement over the cubic scaling of na"ive approaches. We describe how to leverage the properties of rQK matrices for inference in Latent Gaussian Models.
Full work available at URL: https://arxiv.org/abs/1402.4984
Censored data models (62N01) Generalized linear models (logistic models) (62J12) Iterative numerical methods for linear systems (65F10)
Cites Work
- Title not available (Why is that?)
- Title not available (Why is that?)
- Title not available (Why is that?)
- Title not available (Why is that?)
- Gaussian Markov Random Fields
- Fixed Rank Kriging for Very Large Spatial Data Sets
- Functional data analysis.
- Combining Registration and Fitting for Functional Models
- Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations (with discussion)
- Log Gaussian Cox Processes
- Statistical Analysis and Modelling of Spatial Point Patterns
- An Explicit Link between Gaussian Fields and Gaussian Markov Random Fields: The Stochastic Partial Differential Equation Approach
- On the limited memory BFGS method for large scale optimization
- Accurate Approximations for Posterior Moments and Marginal Densities
- Excursion and Contour Uncertainty Regions for Latent Gaussian Models
- Modeling Nonstationary Processes Through Dimension Expansion
- Bayesian Hierarchical Curve Registration
- Hierarchical models for assessing variability among functions
- The Variational Gaussian Approximation Revisited
- Bayesian functional ANOVA modeling using Gaussian process prior distributions
- Bayesian registration of functions and curves
- On Moore-Penrose inverses of quasi-Kronecker structured matrices
Cited In (2)
Uses Software
Recommendations
- Fast covariance estimation for sparse functional data ๐ ๐
- Functional Additive Models ๐ ๐
- Generalized additive models for functional data ๐ ๐
- Functional singular component analysis based functional additive models ๐ ๐
- Fast Multilevel Functional Principal Component Analysis ๐ ๐
- Functional data analysis by matrix completion ๐ ๐
- Fast implementation of partial least squares for function-on-function regression ๐ ๐
- Fast matrix algebra for Bayesian model calibration ๐ ๐
- Fast Univariate Inference for Longitudinal Functional Models ๐ ๐
- Introduction to ``Fast matrix computations for functional additive models by S. Barthelmรฉ ๐ ๐
This page was built for publication: Fast matrix computations for functional additive models
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5963540)