Testing stationarity with time-frequency surrogates

A method is proposed for testing stationarity in an operational sense, i.e., by both including explicitly an observation scale in the definition and elaborating a stationarized reference so as to reject the null hypothesis of stationarity with a controlled level of statistical significance. While the approach is classically based on comparing local vs. global features in the time-frequency plane, the test operates with a family of stationarized surrogates whose analysis allows for a characterization of the null hypothesis. The general principle of the method is outlined, practical issues related to its actual implementation are discussed and a typical example is provided for illustrating the approach and supporting its effectiveness.

[1]  James Theiler,et al.  Testing for nonlinearity in time series: the method of surrogate data , 1992 .

[2]  Nello Cristianini,et al.  An Introduction to Support Vector Machines and Other Kernel-based Learning Methods , 2000 .

[3]  Simon J. Godsill,et al.  Detection of abrupt spectral changes using support vector machines an application to audio signal segmentation , 2002, 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[4]  John B. Shoven,et al.  I , Edinburgh Medical and Surgical Journal.

[5]  W. Martin.  Measuring the degree of non-stationarity by using the Wigner-Ville spectrum , 1984, ICASSP.

[6]  Jun Xiao,et al.  Multitaper time-frequency reassignment , 2006, 2006 14th European Signal Processing Conference.

[7]  Jun Xiao,et al.  Testing Stationarity with Surrogates - A One-Class SVM Approach , 2007, 2007 IEEE/SP 14th Workshop on Statistical Signal Processing.

[8]  Georgios B. Giannakis,et al.  Bibliography on cyclostationarity , 2005, Signal Process..

[9]  M. Basseville.  Distance measures for signal processing and pattern recognition , 1989 .

[10]  Jun Xiao,et al.  Multitaper Time-Frequency Reassignment for Nonstationary Spectrum Estimation and Chirp Enhancement , 2007, IEEE Transactions on Signal Processing.

[11]  C. Doncarli,et al.  Stationarity index for abrupt changes detection in the time-frequency plane , 1996, IEEE Signal Processing Letters.

[12]  Bernhard Schölkopf,et al.  Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond , 2005, IEEE Transactions on Neural Networks.

[13]  Richard G. Baraniuk,et al.  Multiple Window Time Varying Spectrum Estimation , 2000 .

[14]  Richard A. Silverman,et al.  Locally stationary random processes , 2018, IRE Trans. Inf. Theory.

[15]  C. Keylock,et al.  Constrained surrogate time series with preservation of the mean and variance structure. , 2006, Physical review. E, Statistical, nonlinear, and soft matter physics.

[16]  Patrick Flandrin,et al.  Time-Frequency/Time-Scale Analysis , 1998 .

[17]  Paul Honeine,et al.  Time-Frequency Learning Machines , 2007, IEEE Transactions on Signal Processing.

[18]  P. Flandrin,et al.  Detection of changes of signal structure by using the Wigner-Ville spectrum , 1985 .

[19]  Anthony Widjaja,et al.  Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond , 2003, IEEE Transactions on Neural Networks.

[20]  Schreiber,et al.  Improved Surrogate Data for Nonlinearity Tests. , 1996, Physical review letters.

[21]  S. Mallat,et al.  Adaptive covariance estimation of locally stationary processes , 1998 .