On the basis of numerical calculations, we can state that limiting the number of sum elements to 15 is sufficient, as the next element does not change the result by more than 0.05%.
This last summing element adds the output signal issued from the summing element representing the cerebellar nuclei to the signal coding the desired movement.
The efferent copy loops through a second (downstream) summing element (labelled RN in Figure 1B C) which represents the magno-cellular part of the red nucleus.
This inhibitory signal, together with the efferent copy of the neural command (u), is then fed to a summing element which would represent the cerebellar nuclei (labeled CN in Figure 1B,C).
It is noticeable that the motor command (u) reaches both the predictor and this summing element, similar to the messages conveyed by the excitatory mossy fibers that reach both the cerebellar cortex and the cerebellar nuclei.
Thus, the predictor resembles the cerebellar cortex, and the summing element resembles the cerebellar nuclei (Figs. 3, 4).
The same feedback signals reach the predictor and summing element, as the messages conveyed by the mossy fibers reach both the cerebellar cortex and cerebellar nuclei.
To obtain the stochastic matrix, we need to have sums elements of every row of the matrix to sums up to one, but as not all the nodes have out degree edges its inevitable to have rows of all zero which does not sum up to one.
We define the sum of element-by-element squared weights for the training matrices that fall into and Voronoi regions, as (14).
Cost-based abduction, where a numerical weight is assigned to each hypothesis and an optimal solution hypotheses set with minimal sum of element hypotheses' weights is searched, deals with such problems.
