Hyperparameter Optimization of the Perceptron Algorithm for Determining the Feasibility of Research Proposals and Community Service
Higher education in Indonesia includes diploma, bachelor, master, specialist, and doctoral programmes organised by universities. The Institute for Research and Community Service (LPPM) is in charge of assessing lecturers' proposals. This research aims to optimise the Perceptron algorithm to ass...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | Indonesian |
| Published: |
Universitas Lancang Kuning
2024-11-01
|
| Series: | Digital Zone: Jurnal Teknologi Informasi dan Komunikasi |
| Subjects: | |
| Online Access: | https://journal.unilak.ac.id/index.php/dz/article/view/17812 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Higher education in Indonesia includes diploma, bachelor, master, specialist, and doctoral programmes organised by universities. The Institute for Research and Community Service (LPPM) is in charge of assessing lecturers' proposals. This research aims to optimise the Perceptron algorithm to assess proposal eligibility using Turnitin plagiarism scores and reviewer scores. The optimisation results show that Perceptron accuracy reaches 99.44% to 99.63% at various training data ratios. GridSearchCV achieved 100% accuracy, while RandomisedSearchCV recorded accuracy between 98.89% to 99.63%. GridSearchCV also had the lowest MSE , despite higher Loss values, indicating a sacrifice in generalisation ability. Perceptron Default and RandomisedSearchCV had higher MSE and Loss, but remained low. GridSearchCV's AUC reached 100%, while Perceptron Default and RandomisedSearchCV showed very high AUC, ranging from 99.25% to 99.98%. Overall, the Perceptron algorithm is effective in assessing proposal eligibility with high accuracy. |
|---|---|
| ISSN: | 2086-4884 2477-3255 |