Multivariate Self-Similarity: Estimating Self-Similarity Exponents and Applications

TIME: 3:30 PM

LOCATION: GMCS 314

SPEAKER: Charles-Gerard Lucas, Mathematics & Statistics, San Diego State University

ABSTRACT:

Scale invariance is a versatile signal processing paradigm that appears in various real-world applications and can be formalised by self-similarity. However, most practical studies have so far remained univariate, limiting themselves to analyse the different components of the same dataset independently. Yet, the most recent applications often involve the use of many sensors to monitor the same system, for which a relevant study requires a multivariate analysis of the resulting time series. This presentation will cover the modeling and analysis of multivariate self-similarity. An application to epileptic seizure prediction from multi-channel EEG time series will also be presented.

HOST: Jerome Gilles

VIDEO: