Calculating the Effect of Neural Network Parameters on Their Performance

EasyChair Preprint 3298

Abstract

In this research, a three-layer neuronal network was studied and designed, a network capable of learning a set of large data with the help of the reverse propagation error method, and studying the effect of parameter changes (learning step, number of nodes, type of activation dependent for a number of different income signals) and the severe impact This change in the work of the neural network causes it, as the results of these experiments demonstrated the extreme sensitivity of the designed neuron response, which relies on the propagation technique to change these parameters.

Keyphrases: Back propagation algorithm, algorithm instruction, layers, neural networks

@booklet{EasyChair:3298,
  author    = {Manhal Basher Hasan Aga},
  title     = {Calculating the Effect of Neural Network Parameters on Their Performance},
  howpublished = {EasyChair Preprint 3298},
  year      = {EasyChair, 2020}}