Approximand

Each shape affects how well F approximates a given approximand and how quickly an adaptive fuzzy system F approximates when learning tunes the parameters of Aⱼ.

In effect, this result is ideal, as it gives the same rate m^(-(s-1/p)) we could hope to obtain by knowing that the underlying approximand was a ridge function in a specific direction […]

The key task of approximating with granules is seeking for the tightest envelope for the approximand.