Sparsity-driven spatio-temporal EEG source estimation

We consider the problem of sparse source estimation using multiple time samples. We propose the use of a spatio-temporal constraint which is a combination of spatial ℓ
₁
-norm for sparsity and temporal ℓ
₂
-norm for nonspiky pattern in time. We demonstrate the effectiveness of this spatio-temporal method through experiments based on both simulated and human EEG data.