TY - JOUR AU - Alrahhal, Hosam AU - Razan Jamous, PY - 2021/06/08 Y2 - 2024/03/29 TI - A New PSO-based ANN Hyper-parameters SelectionModel JF - WAS Science Nature (WASSN) ISSN: 2766-7715 JA - WASSN VL - 4 IS - 1 SE - Computer Science & Mathematics DO - UR - https://worldascience.com/journals/index.php/wassn/article/view/23 SP - AB - <p><span class="fontstyle0">Finding the best structure of ANN to minimize errors, the processing, and<br>the search time is one of the main objectives in the AI field. In this paper, an<br>enhanced PSO-based selection technique to determine the optimal configuration<br>for the artificial neural network is presented. PSO with 2-D search space is used to<br>select the optimal number of the hidden layer and the number of units per hidden<br>layer. The proposed technique was evaluated using a chemical dataset. The result<br>of testing the proposed technique displayed high prediction accuracy with<br>a minimum error close to zero. In addition, the suggested technique reduces the<br>mean absolute percentage error and mean absolute error significantly compared to<br>ANN and PSO methods. Moreover, the relative error between the expected output<br>and actual target is </span><span class="fontstyle2">approximately between -0.02 and 0.02. </span><span class="fontstyle0">The results of the<br>comparison of the proposed technique with the ANN and PSO showed that the<br>performance of the proposed approach is better in terms of the accuracy of the<br>output prediction.</span> </p> ER -