Classification of functional fragments by regularized linear
classifiers with domain selection

J 2019

Classification of functional fragments by regularized linear classifiers with domain selection

KRAUS, David and Marco STEFANUCCI

Basic information

Original name

Classification of functional fragments by regularized linear classifiers with domain selection

Authors

KRAUS, David (203 Czech Republic, guarantor, belonging to the institution) and Marco STEFANUCCI (380 Italy)

Edition

Biometrika, Oxford, Oxford Univ Press, 2019, 0006-3444

Other information

Language

English

Type of outcome

Článek v odborném periodiku

Field of Study

10103 Statistics and probability

Country of publisher

United Kingdom of Great Britain and Northern Ireland

Confidentiality degree

není předmětem státního či obchodního tajemství

References:

Full Text

Impact factor

Impact factor: 1.632

RIV identification code

RIV/00216224:14310/19:00107190

Organization unit

Faculty of Science

DOI

http://dx.doi.org/10.1093/biomet/asy060

UT WoS

000460615100013

Keywords in English

Classification; Conjugate gradients; Domain selection; Functional data; Partial observation; Regularization; Ridge method

Abstract

V originále

We consider classification of functional data into two groups by linear classifiers based on one-dimensional projections of functions. We reformulate the task of finding the best classifier as an optimization problem and solve it by the conjugate gradient method with early stopping, the principal component method, and the ridge method. We study the empirical version with finite training samples consisting of incomplete functions observed on different subsets of the domain and show that the optimal, possibly zero, misclassification probability can be achieved in the limit along a possibly nonconvergent empirical regularization path. We propose a domain extension and selection procedure that finds the best domain beyond the common observation domain of all curves. In a simulation study we compare the different regularization methods and investigate the performance of domain selection. Our method is illustrated on a medical dataset, where we observe a substantial improvement of classification accuracy due to domain extension.

Links

GJ17-22950Y, research and development project

Name: Statistická inference pro složité náhodné procesy v ekonometrickém modelování

Investor: Czech Science Foundation

Citovat

KRAUS, David and Marco STEFANUCCI. Classification of functional fragments by regularized linear classifiers with domain selection. Biometrika. Oxford: Oxford Univ Press, 2019, vol. 106, No 1, p. 161-180. ISSN 0006-3444. Available from: https://dx.doi.org/10.1093/biomet/asy060.

@article{1468976,
   author = {Kraus, David and Stefanucci, Marco},
   article_location = {Oxford},
   article_number = {1},
   doi = {http://dx.doi.org/10.1093/biomet/asy060},
   keywords = {Classification; Conjugate gradients; Domain selection; Functional data; Partial observation; Regularization; Ridge method},
   language = {eng},
   issn = {0006-3444},
   journal = {Biometrika},
   title = {Classification of functional fragments by regularized linear classifiers with domain selection},
   url = {https://academic.oup.com//biomet/advance-article/doi/10.1093/biomet/asy060/5250873?guestAccessKey=64dbd3b6-5ffc-4dee-8586-3fee35ebded7},
   volume = {106},
   year = {2019}
}

TY  - JOUR
ID  - 1468976
AU  - Kraus, David - Stefanucci, Marco
PY  - 2019
TI  - Classification of functional fragments by regularized linear classifiers with domain selection
JF  - Biometrika
VL  - 106
IS  - 1
SP  - 161-180
EP  - 161-180
PB  - Oxford Univ Press
SN  - 00063444
KW  - Classification
KW  - Conjugate gradients
KW  - Domain selection
KW  - Functional data
KW  - Partial observation
KW  - Regularization
KW  - Ridge method
UR  - https://academic.oup.com//biomet/advance-article/doi/10.1093/biomet/asy060/5250873?guestAccessKey=64dbd3b6-5ffc-4dee-8586-3fee35ebded7
L2  - https://academic.oup.com//biomet/advance-article/doi/10.1093/biomet/asy060/5250873?guestAccessKey=64dbd3b6-5ffc-4dee-8586-3fee35ebded7
N2  - We consider classification of functional data into two groups by linear classifiers based on one-dimensional projections of functions. We reformulate the task of finding the best classifier as an optimization problem and solve it by the conjugate gradient method with early stopping, the principal component method, and the ridge method. We study the empirical version with finite training samples consisting of incomplete functions observed on different subsets of the domain and show that the optimal, possibly zero, misclassification probability can be achieved in the limit along a possibly nonconvergent empirical regularization path. We propose a domain extension and selection procedure that finds the best domain beyond the common observation domain of all curves. In a simulation study we compare the different regularization methods and investigate the performance of domain selection. Our method is illustrated on a medical dataset, where we observe a substantial improvement of classification accuracy due to domain extension.
ER  -

KRAUS, David and Marco STEFANUCCI. Classification of functional fragments by regularized linear classifiers with domain selection. \textit{Biometrika}. Oxford: Oxford Univ Press, 2019, vol.~106, No~1, p.~161-180. ISSN~0006-3444. Available from: https://dx.doi.org/10.1093/biomet/asy060.

Detailed Information on Publication Record