Experimental Analysis of Neural Approaches for Synthetic Angle-of-Attack Estimation

Figure 5

Trade-off of number of neurons, radius, and overlap for GRBF-NN trained with the “original” dataset.

(a) Best training performances with 10, 13, 25, 50, and 100 neurons adopting the best combination of radius and overlap

(b) , , and maximum errors as a function of the radius and overlap parameters for GRBF-NN with neurons