Empirical wavelets: principle, theory and applications

Data driven techniques have been at the center of attention for several years. If supervised techniques have proven their efficacy in many fields, their main drawback is the need for extensive annotated datasets for their training. For certain applications, the availability of such huge datasets is not possible. On the other hand, time/spatial-frequency analysis has been used for decades to characterize signals and images. Data-driven time-frequency analysis techniques have been investigated this last decade. Among them, empirical wavelets have been proven to extract accurate information allowing further analysis. In this talk, we will review the concept of empirical wavelets and define a general mathematical framework. Finally, I will present some applications in signal/image processing.

Bio
https://www.sdsu.edu/experts-directory/jerome-giles