A New PSO-based ANN Hyper-parameters SelectionModel


  • Dr. Hosam Alrahhal Faculty of Engineering, NAHDA University, EGYPT.
  • Prof. Razan Jamous Faculty of Engineering and Applied Science, University of Regina, Regina, Canada


Particle swarm optimization, Artificial Neural Network, Hyperparameter, hidden layer.


Finding the best structure of ANN to minimize errors, the processing, and
the search time is one of the main objectives in the AI field. In this paper, an
enhanced PSO-based selection technique to determine the optimal configuration
for the artificial neural network is presented. PSO with 2-D search space is used to
select the optimal number of the hidden layer and the number of units per hidden
layer. The proposed technique was evaluated using a chemical dataset. The result
of testing the proposed technique displayed high prediction accuracy with
a minimum error close to zero. In addition, the suggested technique reduces the
mean absolute percentage error and mean absolute error significantly compared to
ANN and PSO methods. Moreover, the relative error between the expected output
and actual target is approximately between -0.02 and 0.02. The results of the
comparison of the proposed technique with the ANN and PSO showed that the
performance of the proposed approach is better in terms of the accuracy of the
output prediction.




How to Cite

Alrahhal, H., & Razan Jamous. (2021). A New PSO-based ANN Hyper-parameters SelectionModel. WAS Science Nature (WASSN) ISSN: 2766-7715, 4(1). Retrieved from https://worldascience.com/journals/index.php/wassn/article/view/23



Computer Science & Mathematics