SLANTS: Sequential Adaptive Nonlinear Modeling of Time Series

Qiuyi Han, Jie Ding, Edoardo M. Airoldi, Vahid Tarokh

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

We propose a method for adaptive nonlinear sequential modeling of time series data. Data are modeled as a nonlinear function of past values corrupted by noise, and the underlying nonlinear function is assumed to be approximately expandable in a spline basis. We cast the modeling of data as finding a good fit representation in the linear span of multidimensional spline basis, and use a variant of $l-1$-penalty regularization in order to reduce the dimensionality of representation. Using adaptive filtering techniques, we design our online algorithm to automatically tune the underlying parameters based on the minimization of the regularized sequential prediction error. We demonstrate the generality and flexibility of the proposed approach on both synthetic and real-world datasets. Moreover, we analytically investigate the performance of our algorithm by obtaining both bounds on prediction errors and consistency in variable selection.

Original languageEnglish (US)
Article number7950945
Pages (from-to)4994-5005
Number of pages12
JournalIEEE Transactions on Signal Processing
Volume65
Issue number19
DOIs
StatePublished - Oct 1 2017

Fingerprint

Splines
Time series
Adaptive filtering

Keywords

  • Adaptive filtering
  • Data prediction
  • Group LASSO
  • Nonlinearity
  • SLANTS
  • Sequential modeling
  • Spline
  • Time series

Cite this

SLANTS : Sequential Adaptive Nonlinear Modeling of Time Series. / Han, Qiuyi; Ding, Jie; Airoldi, Edoardo M.; Tarokh, Vahid.

In: IEEE Transactions on Signal Processing, Vol. 65, No. 19, 7950945, 01.10.2017, p. 4994-5005.

Research output: Contribution to journalArticle

Han, Qiuyi ; Ding, Jie ; Airoldi, Edoardo M. ; Tarokh, Vahid. / SLANTS : Sequential Adaptive Nonlinear Modeling of Time Series. In: IEEE Transactions on Signal Processing. 2017 ; Vol. 65, No. 19. pp. 4994-5005.
@article{8d2e3ecf20064dfda737aaf9a669bbc3,
title = "SLANTS: Sequential Adaptive Nonlinear Modeling of Time Series",
abstract = "We propose a method for adaptive nonlinear sequential modeling of time series data. Data are modeled as a nonlinear function of past values corrupted by noise, and the underlying nonlinear function is assumed to be approximately expandable in a spline basis. We cast the modeling of data as finding a good fit representation in the linear span of multidimensional spline basis, and use a variant of $l-1$-penalty regularization in order to reduce the dimensionality of representation. Using adaptive filtering techniques, we design our online algorithm to automatically tune the underlying parameters based on the minimization of the regularized sequential prediction error. We demonstrate the generality and flexibility of the proposed approach on both synthetic and real-world datasets. Moreover, we analytically investigate the performance of our algorithm by obtaining both bounds on prediction errors and consistency in variable selection.",
keywords = "Adaptive filtering, Data prediction, Group LASSO, Nonlinearity, SLANTS, Sequential modeling, Spline, Time series",
author = "Qiuyi Han and Jie Ding and Airoldi, {Edoardo M.} and Vahid Tarokh",
year = "2017",
month = "10",
day = "1",
doi = "10.1109/TSP.2017.2716898",
language = "English (US)",
volume = "65",
pages = "4994--5005",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "19",

}

TY - JOUR

T1 - SLANTS

T2 - Sequential Adaptive Nonlinear Modeling of Time Series

AU - Han, Qiuyi

AU - Ding, Jie

AU - Airoldi, Edoardo M.

AU - Tarokh, Vahid

PY - 2017/10/1

Y1 - 2017/10/1

N2 - We propose a method for adaptive nonlinear sequential modeling of time series data. Data are modeled as a nonlinear function of past values corrupted by noise, and the underlying nonlinear function is assumed to be approximately expandable in a spline basis. We cast the modeling of data as finding a good fit representation in the linear span of multidimensional spline basis, and use a variant of $l-1$-penalty regularization in order to reduce the dimensionality of representation. Using adaptive filtering techniques, we design our online algorithm to automatically tune the underlying parameters based on the minimization of the regularized sequential prediction error. We demonstrate the generality and flexibility of the proposed approach on both synthetic and real-world datasets. Moreover, we analytically investigate the performance of our algorithm by obtaining both bounds on prediction errors and consistency in variable selection.

AB - We propose a method for adaptive nonlinear sequential modeling of time series data. Data are modeled as a nonlinear function of past values corrupted by noise, and the underlying nonlinear function is assumed to be approximately expandable in a spline basis. We cast the modeling of data as finding a good fit representation in the linear span of multidimensional spline basis, and use a variant of $l-1$-penalty regularization in order to reduce the dimensionality of representation. Using adaptive filtering techniques, we design our online algorithm to automatically tune the underlying parameters based on the minimization of the regularized sequential prediction error. We demonstrate the generality and flexibility of the proposed approach on both synthetic and real-world datasets. Moreover, we analytically investigate the performance of our algorithm by obtaining both bounds on prediction errors and consistency in variable selection.

KW - Adaptive filtering

KW - Data prediction

KW - Group LASSO

KW - Nonlinearity

KW - SLANTS

KW - Sequential modeling

KW - Spline

KW - Time series

UR - http://www.scopus.com/inward/record.url?scp=85021816947&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021816947&partnerID=8YFLogxK

U2 - 10.1109/TSP.2017.2716898

DO - 10.1109/TSP.2017.2716898

M3 - Article

AN - SCOPUS:85021816947

VL - 65

SP - 4994

EP - 5005

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 19

M1 - 7950945

ER -