151. Performance analysis of scaled conjugate gradient (SCG) algorithm on computing problems.
- Author
-
Ariana, A. A. G. B., Wisky, Irzal Arief, Ginantra, Ni Luh Wiwik Sri Rahayu, Firmansyah, Moch. Rachmandany, and Daengs, G. S. Achmad
- Subjects
- *
ALGORITHMS , *PROBLEM solving , *ARTIFICIAL neural networks - Abstract
The artificial neural network has several training functions that can speed up the training process of the standard backpropagation algorithm. Therefore, the purpose of this research is to evaluate the scaled conjugate gradient algorithm's capability and performance, which develops the training function of standard backpropagation to solve computational problems. The dataset used in this paper uses quantitative data from export data of jewelry and valuable goods by the leading destination country, which is processed from documents customs of the Directorate General of Customs and Excise (PEB and PIB) and quoted from the Indonesian Statistical Publication. A network architecture model will be formed and determined based on this data, including 7-7-1, 7-14-1, and 7-21-1. Based on these three models after training and testing, the results show that the model with the best performance and accuracy is 7-14-1 with a performance value of 0.001118426, the lowest among the three other models and an accuracy of 90.9% (higher than the two models other). So it can be concluded that the SCG algorithm with the 7-14-1 model can be used to solve computational problems, as evidenced by the best performance and accuracy values. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF