Parabola As an Activation Function of Artificial Neural Networks

The use of the parabola and its branches as a nonlinearity expanding the logical capabilities of artificial neurons is considered. In particular, the applicability of parabola branches to the construction of an s-shaped function is suitable for tuning a neural network through reverse error propagation is determined. Solutions to typical problem of function XOR construction are shown using a rotated parabola. The main focus of modern research is to reduce computational complexity or, on the contrary, accelerate calculations by parallelizing a nonlinear function, i.e. by hardware redundancy.