On the Use of Reproducing Kernel Hilbert Spaces in Functional Classification

DOI10.1080/01621459.2017.1320287zbMATH Open1402.68152arXiv1507.04398OpenAlexW2485393623MaRDI QIDQ4559702FDOQ4559702

Authors: José R. Berrendero, José L. Torrecilla, Antonio Cuevas 

Publication date: 4 December 2018

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: The H'ajek-Feldman dichotomy establishes that two Gaussian measures are either mutually absolutely continuous with respect to each other (and hence there is a Radon-Nikodym density for each measure with respect to the other one) or mutually singular. Unlike the case of finite dimensional Gaussian measures, there are non-trivial examples of both situations when dealing with Gaussian stochastic processes. This paper provides: (a) Explicit expressions for the optimal (Bayes) rule and the minimal classification error probability in several relevant problems of supervised binary classification of mutually absolutely continuous Gaussian processes. The approach relies on some classical results in the theory of Reproducing Kernel Hilbert Spaces (RKHS). (b) An interpretation, in terms of mutual singularity, for the "near perfect classification" phenomenon described by Delaigle and Hall (2012). We show that the asymptotically optimal rule proposed by these authors can be identified with the sequence of optimal rules for an approximating sequence of classification problems in the absolutely continuous case. (c) A new model-based method for variable selection in binary classification problems, which arises in a very natural way from the explicit knowledge of the RN-derivatives and the underlying RKHS structure. Different classifiers might be used from the selected variables. In particular, the classical, linear finite-dimensional Fisher rule turns out to be consistent under some standard conditions on the underlying functional model.

Full work available at URL: https://arxiv.org/abs/1507.04398

Recommendations

• On functional reproducing kernels
• Approximating functions in reproducing kernel Hilbert spaces via statistical learning theory
• Functional reproducing kernel Hilbert spaces for non-point-evaluation functional data
• Soft and hard classification by reproducing kernel Hilbert space methods
• Functional Classification in Hilbert Spaces
• On the kernel rule for function classification
• Some properties of Gaussian reproducing kernel Hilbert spaces and their implications for function approximation and learning theory
• On the theory of reproducing kernel Hilbert spaces
• A reproducing kernel Hilbert space approach to functional linear regression
• Regression models for functional data by reproducing kernel Hilbert spaces methods

zbMATH Keywords

variable selectionRadon-Nikodym derivativesabsolutely continuitysupervised functional classificationmutually singular processes

Mathematics Subject Classification ID

Nonparametric regression and quantile regression (62G08) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Learning and adaptive systems in artificial intelligence (68T05) Hilbert spaces with reproducing kernels (= (proper) functional Hilbert spaces, including de Branges-Rovnyak and other structured spaces) (46E22)

Cites Work

• Nonparametric functional data analysis. Theory and practice.
• Variable selection in functional data classification: a maxima-hunting proposal
• A partial overview of the theory of statistics with functional data
• Inference for functional data with applications
• Variable selection in infinite-dimensional problems
• Modern Multivariate Statistical Techniques
• Methodology and theory for partial least squares applied to functional data
• Title not available (Why is that?)
• Classification methods for Hilbert data based on surrogate density
• Brownian motion. With an appendix by Oded Schramm and Wendelin Werner
• Title not available (Why is that?)
• Achieving near Perfect Classification for Functional Data
• Radon-Nikodym Derivatives of Gaussian Measures
• Supervised classification for a family of Gaussian functional models
• Supervised classification of diffusion paths
• On equivalence of Gaussian measures
• Logistic regression with Brownian-like predictors
• An Approach to Time Series Analysis
• Extraction and Detection Problems and Reproducing Kernel Hilbert Spaces
• Radon-Nikodym derivatives with respect to measures induced by discontinuous independent-increment processes
• Equivalence and perpendicularity of Gaussian processes
• Componentwise classification and clustering of functional data
• On Gaussian Measures Equivalent to Wiener Measure
• The mRMR variable selection method: a comparative study for functional data

Cited In (21)

• Optimal Linear Discriminant Analysis for High-Dimensional Functional Data
• On functional logistic regression: some conceptual issues
• Supervised classification of geometrical objects by integrating currents and functional data analysis
• An adaptive group Lasso approach for domain selection in functional generalized linear models
• Optimal rate for prediction when predictor and response are functions
• Deep neural network classifier for multidimensional functional data
• On optimal regression trees to detect critical intervals for multivariate functional data
• Fast convergence on perfect classification for functional data
• Optimal classification for functional data
• Review on functional data classification
• Optimal classification of Gaussian processes in homo- and heteroscedastic settings
• Functional Classification in Hilbert Spaces
• A Reproducing Kernel Hilbert Space Framework for Functional Classification
• Supervised classification for a family of Gaussian functional models
• Variable selection in competing risks models based on quantile regression
• Multiclass classification for multidimensional functional data through deep neural networks
• Functional data analysis: an introduction and recent developments
• On Mahalanobis distance in functional settings
• The optimal rate of canonical correlation analysis for stochastic processes
• Covariance estimation error of incomplete functional data under RKHS framework
• On functional reproducing kernels

This page was built for publication: On the Use of Reproducing Kernel Hilbert Spaces in Functional Classification

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4559702)