FIGURE 9.

Test Case 1: (A) original parameters α ⁱ plotted against the corresponding learned parameters ${\widehat{\alpha }}^i$ for the training individuals (i = 1, …, N _P), for different noise magnitudes σ _offline and (B) overall L ² relative prediction errors obtained for different values of σ _offline and σ _online. In (B), each black cross represent a testing individual, the blue curve represent the arithmetic mean, the red curve the geometric mean