|
|
 |
| |
|
|
|
49 Amplified Generalized Fractional Neural Network Approximations at a Reduced Finite Domain |
|
|
|
PP: 489-503 |
|
|
doi:10.18576/pfda/120301
|
|
|
|
Author(s) |
|
|
|
George A. Anastassiou,
|
|
|
|
Abstract |
|
|
| This research deals with the determination of the rate of generalized fractional pointwise, uniform and Lp, p ≥ 1, convergences to the unit operator of 49 specific ”normalized cusp neural network operators”. The cusp is a compact support activation function, which derives from the composition of two specific activation functions having as domain the whole real line. We employ 7 different known activation functions. These convergences are given via the moduli of continuity of the right and left generalized fractional derivatives of Caputo type of the engaged function in the form of Jackson type inequalities. The composition of activation functions aims to more flexible and powerful neural networks utilizing the reduction of infinite domains to the one domain of compact support. |
|
|
|
|
|
|
|
