Testing Stationarity with Surrogates — A One-Class SVM Approach

Jun Xiao (1) , Pierre Borgnat (1) , Patrick Flandrin (1) , Cédric Richard (2)

1 Phys-ENS - Laboratoire de Physique de l'ENS Lyon
2 LM2S - Laboratoire Modélisation et Sûreté des Systèmes

An operational framework is developed for testing stationarity relatively to an observation scale, in both stochastic and deterministic contexts. The proposed method is based on a comparison between global and local time-frequency features. The originality is to make use of a family of stationary surrogates for defining the null hypothesis and to base on them a statistical test implemented as a one-class Support Vector Machine. The time-frequency features extracted from the surrogates are considered as a learning set and used to detect departure from stationnarity. The principle of the method is presented, and some results are shown on typical models of signals that can be thought of as stationary or nonstationary, depending on the observation scale used.

Keywords: One-Class Classification; Time-Frequency Analysis; Support Vector Machines; Stationarity Test

Domains: Computer Science [cs]; Signal and Image Processing; Engineering Sciences [physics]; Signal and Image processing

Jun Xiao, Pierre Borgnat, Patrick Flandrin, Cédric Richard. Testing Stationarity with Surrogates — A One-Class SVM Approach. 2007 IEEE/SP 14th Statistical Signal Processing Workshop (SSP '07), IEEE/SP, Aug 2007, Madison, Wisconsin, United States. pp.720-724. ⟨ensl-00175481⟩