Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Prosiding Seminar Nasional Official Statistics
Muhammad Yusuf Aristyanto
Unknown Affiliation
Humanities Computer Science & IT Economics, Econometrics & Finance Social Sciences

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Pengembangan Metode Neural Machine Translation Berdasarkan Hyperparameter Neural Network Muhammad Yusuf Aristyanto; Robert Kurniawan
Seminar Nasional Official Statistics Vol 2021 No 1 (2021): Seminar Nasional Official Statistics 2021
Publisher : Politeknik Statistika STIS

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (532.299 KB) | DOI: 10.34123/semnasoffstat.v2021i1.789

Abstract

Humans as social beings who always want to be in touch with other human’s force humans to communicate with each other. This is where the role of language becomes very important, because with language, it will be easy to understand what other people want to convey. For this reason, it is necessary to have a media that can help to understand the various languages in the world, one of which is machine translation. One method that can be used to make machine translators is Neural Machine Translation (NMT). The existing NMT still has various shortcomings and needs to be further developed. Among them is the problem of overfitting which makes the model less able to generalize to other data being tested. Many things affect the performance of the NMT, one of which is the size of the hyperparameters used and the model architecture used. However, there is no definite measure that can be used to produce a model with the best performance. So, this study aims to develop the NMT model architecture and perform simulations on each Neural Network hyperparameter and the size of the model architecture, including batch size, epoch, optimizer, activation function, and dropout rate. The results obtained are the development model can overcome the overfitting problem of the previous model with an accuracy of 72.24% and a BLEU score of 45.83% which was carried out on other test data.
Co-Authors Robert Kurniawan, Robert