Seminar: Revisiting and Testing Stationarity
Speaker: Patrick Flandrin, CNRS Senior Researcher, Physics Department, Ecole Normale Supérieure de Lyon, France
Date: October 26, 2011
Time: 1:10 p.m.
Location: 3043 ECpE Building Addition
Abstract: Albeit ubiquitous as an assumption or as a pre-requisite in many signal processing tasks, the concept of stationarity is often implicitly understood in a loose sense. The purpose of this lecture is to revisit it from an operational perspective that explicitly takes into account the observation scale. A general, time-frequency-based, framework is described for testing such a relative stationarity via the introduction of stationarized surrogate data. Two variations are discussed, based on either dissimilarity measures between local and global spectra, or machine learning approaches. Different extensions, including wavelet-based tests for images and transient detection, are considered.
Speaker bio: Patrick Flandrin (F) received the engineer degree from ICPI Lyon, France, in 1978, and the Doct.-Ing. and Docteur d’Etat degrees from INP Grenoble, France, in 1982 and 1987, respectively. He joined CNRS in 1982, where he is currently a Research Director. Since 1991, he has been with the Signals, Systems, and Physics Group, Physics Department, Ecole Normale Supérieure de Lyon, France.
Prof. Flandrin has been a major contributor to the theory of (bilinear) Time-Frequency representations and non-stationary signal analysis. He played a major role in the developments of the wavelet theory and the analysis of fractional Brownian motion. Recently, he opened a new research direction studying the Empirical Mode Decomposition and revisiting stationarity with significant contributions on stationarity tests.
Prof. Flandrin is author of the book titled, Time-Frequency/Time-Scale Analysis and has authored more than 250 journal and conference proceeding research articles.
Prof. Flandrin received several research awards including Philip Morris Prize in Mathematics (1991); SPIE Wavelet Pioneer Award (2001); “Prix Michel Monpetit” from the French Academy of Sciences (2001); and Silver Medal from CNRS (2010).
Prof. Flandrin has served as a guest co-editor of the Special Issue “Wavelets and Signal Processing” of the IEEE Transactions on Signal Processing (1993); Technical Program Chairman, IEEE-International Symposium on Time-Frequency and Time-Scale Analysis (1994); and Program Chairman, French GRETSI Symposium on Signal and Image Processing, every two years since 2001. He has been Associate Editor, IEEE Transactions on Signal Processing (1990-93, 2008-present); and Associate Editor, EURASIP Signal Processing (1994-05). Prof. Flandrin is currently on the Editorial Board of Applied and Computational Harmonic Analysis, The Journal of Fourier Analysis and Applications, Signal Image and Video Processing, and Advances in Adaptive Data Analysis. He has also been Member, IEEE Signal Processing Theory and Methods Technical Committee (1993-04). Prof. Flandrin spent one semester in Cambridge, UK, as an invited long-term resident of the Isaac Newton Institute for Mathematical Sciences (1998). Finally, Prof. Flandrin is a Fellow of the IEEE (2002) and of EURASIP (2009).