The Epsilon-Skew-Normal dictionary for the decomposition of single- and multichannel biomedical recordings using Matching Pursuit algorithms

Matching Pursuit based algorithms are a well-established method for decomposing single- and multichannel biomedical recordings. Due to the time-frequency-characteristics, Gabor dictionaries are commonly used. However, symmetric Gabor atoms fail at approximating asymmetric oscillatory components. We present the Epsilon-Skew-Normal dictionary which is built from symmetric as well as asymmetric components. The new dictionary can be considered as an extension of the Gabor dictionary. We compared both dictionaries based on the decomposition of simulated as well as real EEG data and found that the Epsilon-Skew-Normal dictionary causes smaller decomposition errors compared to the Gabor dictionary.