showing the results for yp, =0.02, or a memory depth of 50 samples, where the system

detects a switch back to regime 2 at the end of inspirations in regime 1, before switching

immediately back to regime 1), or even no switches at all (in Figure 4-6, with Ye/ =0.03

or a memory depth of 33 samples, the mandatory breaths are barely detected, because the

criterion values are smoothed less because of lesser samples for the average).


Figure 4-5 Results: (a) Flow: desired in blue, predicted in red; (b) MSE criterion, full line
 is for expert 1, dotted line is for expert 2; (c) Winner; (d) Pressure.

 Also with only a memory depth of 33 samples, the two experts compete in the

expiration phase without memory of what kind of inspiration has just taken place, and

expert 2 seems to model that phase better.

 These examples show the effect of varying the memory depth of the criterion, but

from now on, the same value of Yep =0.01 will be used for all simulations, for it ensures


no unwarranted switches in all cases.


 0 0 -' I I
 500 1000 1500 2000 2500 3000

1 F 1II I n I I I I