Skip to main content
. Author manuscript; available in PMC: 2020 Dec 4.
Published in final edited form as: Neuroimage. 2020 Jan 13;211:116528. doi: 10.1016/j.neuroimage.2020.116528

Figure 1:

Figure 1:

Mixed-norm penalty term P2,1,1(Θ) for regularizing the loss function. The penalty term is constructed by first isolating all 3D Gabor coefficient vectors across the dictionary elements and space, and then aggregating their 2 norm. As a result, it promotes sparsity in space and Gabor coefficients, while being invariant to the orientation of the dipole currents.