Skip to main content
. 2020 Mar 9;9:e45743. doi: 10.7554/eLife.45743

Figure 11. Iterative fitting of subunits, partitioned into four steps.

Figure 11.

The subunit kernels (Kn) and weights (wn) are randomly initialized, and used to compute soft cluster assignments (αn,t upper left), followed by cluster centroid computation (Cn - upper right), estimation of subunit kernels (Kn - lower right) and subunit weights (wn - lower left). The summations are only over times when the cell generated a spike (tsp).