Claim Missing Document
Check
Articles

Found 8 Documents
Search

Sistem Identifikasi Pembicara Berbahasa Indonesia Menggunakan X-Vector Embedding Misbullah, Alim; Saifullah Sani, Muhammad; Husaini; Farsiah, Laina; Zahnur; Martiwi Sukiakhy, Kikye
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 11 No 2: April 2024
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25126/jtiik.20241127866

Abstract

Penyemat pembicara adalah vektor yang terbukti efektif dalam merepresentasikan karakteristik pembicara sehingga menghasilkan akurasi yang tinggi dalam ranah pengenalan pembicara. Penelitian ini berfokus pada penerapan x-vectors sebagai penyemat pembicara pada sistem identifikasi pembicara berbahasa Indonesia yang menggunakan model speaker identification. Model dibangun dengan menggunakan dataset VoxCeleb sebagai data latih dan dataset INF19 sebagai data uji yang dikumpulkan dari suara mahasiswa dan mahasiswi Informatika Universitas Syiah Kuala angkatan 2019. Fitur-fitur yang digunakan diekstrak dari dataset audio dengan menggunakan dua jenis konfigurasi mel frequency cepstral coefficients (MFCC). Untuk membangun model, fitur-fitur diekstrak dengan menggunakan MFCC, dihitung voice activity detection (VAD), dilakukan augmentasi dan normalisasi fitur menggunakan cepstral mean and variance normalization (CMVN) serta dilakukan filtering. Sedangkan proses pengujian model hanya membutuhkan fitur-fitur yang diekstrak dengan menggunakan MFCC dan dihitung VAD. Selanjutnya, dibangun empat model dengan cara mengombinasikan dua jenis konfigurasi MFCC dan dua jenis arsitektur Deep Neural Network (DNN) yang memanfaatkan Time Delay Neural Network (TDNN). Model terbaik dipilih berdasarkan akurasi tertinggi yang dihitung menggunakan metrik equal error rate (EER) dan durasi ekstraksi x-vectors tersingkat dari keempat model. Nilai EER dari model yang terbaik untuk dataset VoxCeleb1 bagian test sebesar 3,51%, inf19_test_td sebesar 1,3%, dan inf19_test_tid sebesar 1,4%. Durasi ekstraksi x-vectors menggunakan model terbaik untuk dataset data train berdurasi 6 jam 42 menit 39 detik, VoxCeleb1 bagian test berdurasi 2 menit 24 detik, inf19_enroll berdurasi 18 detik, inf19_test_td berdurasi 25 detik, dan inf19_test_tid berdurasi 9 detik. Arsitektur DNN kedua dan konfigurasi MFCC kedua yang telah dirancang menghasilkan model yang lebih kecil, akurasi yang lebih baik terutama untuk dataset pembicara berbahasa Indonesia, dan durasi ekstraksi x-vectors yang lebih singkat.
Performance Assessment of Machine Learning and Transformer Models for Indonesian Multi-Label Hate Speech Detection Bagestra, Ricky; Misbullah, Alim; Zulfan, Zulfan; Rasudin, Rasudin; Farsiah, Laina; Nazhifah, Sri Azizah
Infolitika Journal of Data Science Vol. 2 No. 2 (2024): November 2024
Publisher : Heca Sentra Analitika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.60084/ijds.v2i2.235

Abstract

Hate speech, characterized by language that incites discrimination, hostility, or violence against individuals or groups based on attributes such as race, religion, or gender, has become a critical issue on social media platforms. In Indonesia, unique linguistic complexities, such as slang, informal expressions, and code-switching, complicate its detection. This study evaluates the performance of Support Vector Machine (SVM), Naive Bayes, and IndoBERT models for multi-label hate speech detection on a dataset of 13,169 annotated Indonesian tweets. The results show that IndoBERT outperforms SVM and Naive Bayes across all metrics, achieving an accuracy of 93%, F1-score of 91%, precision of 91%, and recall of 91%. IndoBERT's contextual embeddings effectively capture nuanced relationships and complex linguistic patterns, offering superior performance in comparison to traditional methods. The study addresses dataset imbalance using BERT-based data augmentation, leading to significant metric improvements, particularly for SVM and Naive Bayes. Preprocessing steps proved essential in standardizing the dataset for effective model training. This research underscores IndoBERT's potential for advancing hate speech detection in non-English, low-resource languages. The findings contribute to the development of scalable, language-specific solutions for managing harmful online content, promoting safer and more inclusive digital environments.
Analisis Performa Segmentasi Citra MRI Tumor Otak dengan Arsitektur U-Net dan Res-UNet Misbullah, Alim; Mursyida, Waliam; Farsiah, Laina; Nazaruddin, Nazaruddin; Sukiakhy, Kikye Martiwi; Husaini, Husaini; Basrul, Basrul
J-SIGN (Journal of Informatics, Information System, and Artificial Intelligence) Vol 2, No 02 (2024): November
Publisher : Department of Informatics, Universitas Syiah Kuala

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24815/j-sign.v2i02.41358

Abstract

Diagnosis tumor otak melalui MRI menghadapi tantangan akibat keterbatasan dalam visualisasi morfologi, lokasi, dan batas-batas tumor. Format MRI yang biasanya dua dimensi memerlukan interpretasi manual oleh radiolog, yang meningkatkan risiko kesalahan manusia. Untuk meningkatkan akurasi segmentasi MRI, pendekatan pembelajaran mendalam seperti Convolutional Neural Networks (CNN) telah diterapkan untuk menyoroti area-area penting. Studi ini membandingkan dua arsitektur CNN, U-Net dan Res-UNet, untuk segmentasi tumor otak menggunakan dataset Brain Tumor Segmentation Challenge (BraTS) 2020. Kedua model dilatih dengan pengaturan yang serupa dan dievaluasi berdasarkan kemampuannya mengidentifikasi area kunci, termasuk inti tumor, edema, dan area tumor yang mengalami peningkatan. Model ini menggunakan optimizer Adam dan fungsi loss categorical crossentropy, dengan metrik evaluasi termasuk akurasi. Hasil menunjukkan bahwa U-Net mencapai performa optimal pada 35 epoch dengan ukuran batch 64 dan learning rate 0,001, menghasilkan nilai loss terendah (0,0140) dan akurasi tertinggi (99,5%). Meskipun Res-UNet juga mencapai akurasi tinggi (99,3%), nilai loss yang lebih tinggi menunjukkan bahwa model ini kurang efektif dibandingkan U-Net.
A Threshold-based Cloud Resource Allocation Framework with Quality of Services Considerations Husaini, Husaini; Misbullah, Alim; Farsiah, Laina
Transcendent Journal of Mathematics and Applications Vol 2, No 1 (2023)
Publisher : Syiah Kuala University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24815/tjoma.v2i1.31694

Abstract

Allocating the number of resources needed by cloud applications is very crucial concern in the cloud environment. If the resource allocation is not managed precisely, the cloud services may starve during the peak load time or waste the resources during the off-peak time. Auto-scaling mechanism is one approach used in cloud environment in which service providers can maintain the resources and reduce waste resources by automatically increasing or decreasing them when needed. It is still difficult to predict the client-side experience which later will cause in decreasing performance because of lacking computing instances. This paper focuses on allocating resources at the application level for the efficient resource utilization and presents a novel cloud resource management framework. The proposed system monitored the end-users response time directly from client-side. Several thresholds were defined with Quality of Services (QoS) considerations which include response time and error rates sampling to optimize the decision of reallocating the virtual resources. The results dynamically allocate the virtual resources among the cloud applications based on their workload. Based on the experimental results, the recommendation threshold is 0.6 for the cloud system, as it can improve performance while minimizing costs.
SISTEM REKOMENDASI PEMILIHAN PROGRAM STUDI BERBASIS HYBRID MENGGUNAKAN PENDEKATAN DEEP LEARNING Misbullah, Alim; Akbar, Mufid; Nazaruddin, Nazaruddin; Farsiah, Laina; Husaini, Husaini; Zulfan, Zulfan
CYBERSPACE: Jurnal Pendidikan Teknologi Informasi Vol 9, No 1 (2025)
Publisher : UIN Ar-Raniry

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22373/cj.v9i1.28944

Abstract

Education plays a critical role in shaping career decisions for the future. However, many students encounter difficulties in selecting suitable academic programs, often stemming from a lack of confidence in their ability to make appropriate decisions. Consequently, students may choose study programs that do not align with their personal characteristics. This study emphasizes the importance of providing comprehensive information about various academic programs offered in higher education and developing tools to assist prospective students in making informed decisions. To address these challenges, a recommendation system using Hybrid Filtering technology has been developed. The system integrates Content-Based Filtering and Collaborative Filtering methods within the TensorFlow Recommenders System (TFRS) framework. The study utilized data from undergraduate students of the Faculty of Mathematics and Natural Sciences (FMIPA) across seven academic programs. By employing 10 features representing students' interests and talents, the recommendation system generated accurate and tailored suggestions for study programs. The model was trained and evaluated using both real and augmented (augmented) datasets with predefined hyperparameters. Results demonstrated that using only the real dataset achieved a Top-1 accuracy of 0.59 and a Top-5 accuracy of 0.97. When incorporating the augmented dataset, the Top-1 accuracy improved to 0.66, while the Top-5 accuracy reached 1.0. The findings reveal that combining real and augmented datasets enhances average accuracy by approximately 10% compared to using the real dataset alone. Additionally, the study program recommendations produced by the model showed significant improvement in quality. A web-based recommendation system utilizing the TFRS model was developed and positively evaluated by FMIPA students. User feedback indicated high satisfaction with the system's recommendations, demonstrating its effectiveness in guiding students toward suitable academic programs.
Penerapan Metode Human Centered Design Pada Perancangan Sistem Pengaduan Masyarakat Desa Berbasis Website Misbullah, Alim; Nazaruddin, Nazaruddin; Asma Liza, Lia; Rasudin, Rasudin; Martiwi Sukiakhy, Kikye; Muzailin, Muzailin; Farsiah, Laina; Zulfan, Zulfan
J-SIGN (Journal of Informatics, Information System, and Artificial Intelligence) Vol 1, No 02 (2023): November
Publisher : Department of Informatics, Universitas Syiah Kuala

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24815/j-sign.v1i2.35111

Abstract

Selama ini, pengaduan masyarakat yang ingin disampaikan secara lisan pada tingkat pedesaan biasanya harus menunggu waktu yang tepat untuk bertemu aparatur desa. Selain itu, pengaduan yang telah disampaikan melalui media komunikasi daring seringnya menyulitkan apratur desa untuk menyimpan dan mengorganisir data tersebut. Pada penelitian ini, sistem informasi berbasis website akan diimplementasikan dalam membangun sistem pengaduan masyarakat desa yang dapat diakses secara daring untuk melaporkan masalah yang terjadi. Sistem pengaduan masyarakat desa dibangun dengan menerapkan metode Human Centered Design yang terdiri dari beberapa tahapan yaitu: spesifikasi konteks pengguna, spesifikasi kebutuhan pengguna, merancang sistem, dan pengujian sistem. Pengujian kualitas sistem dilakukan dengan menggunakan 2 (dua) metode diantaranya blackbox dan webqual. Dengan menggunakan webqual, pengujian sistem pengaduan masyarakat desa mendapatkan hasil yang tinggi pada skala 5 yaitu 67,19% untuk kegunaan, 64,58% untuk kualitas informasi, 62,92% untuk interaksi pengguna, dan 65,04% untuk kualitas website. Hasil pengujian sistem tersebut dapat menjadi pertimbangan awal untuk memutuskan penggunaan sistem pengaduan masyarakat desa nantinya.
Penerapan Aplikasi-Aplikasi Microsoft Office dan Google Docs dalam Upaya Peningkatan Media Pembelajaran di Madrasah Aliyah Negeri 5 Bireuen Irvanizam, Irvanizam; Misbullah, Alim; Zulfan, Zulfan; Farsiah, Laina; Subianto, Muhammad
PESARE: Jurnal Pengabdian Sains dan Rekayasa Vol 1, No 1 (2023): Oktober 2023
Publisher : Universitas Syiah Kuala

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24815/pesare.v1i1.33833

Abstract

This community service activity aims to introduce Microsoft Office and Google Docs applications to teachers and students at Madrasah Aliyah Negeri (MAN) 5 Bireuen as an online teaching media in performing teaching and learning processes during the COVID-19 pandemic. This activity was carried out by a community service team of lecturers from the Department of Informatics, Universitas Syiah Kuala. The activity was held for three days from 2 until 4 April 2021 and consisted of two sessions. The first session introduced Microsoft Office applications for learning at the high school level. The second session demonstrates Google Docs applications for providing teaching materials. The activity participants were very enthusiastic about participating in this activity by asking lots of questions and being explained by the community service team. The result of this activity is that teachers find it very easy and quick to understand how to use these applications for their teaching and learning activities. They hope that online learning activities using the website-based Content Management System method will continue to be carried out as future works.
SISTEM REKOMENDASI PEMILIHAN PROGRAM STUDI BERBASIS HYBRID MENGGUNAKAN PENDEKATAN DEEP LEARNING Misbullah, Alim; Akbar, Mufid; Nazaruddin, Nazaruddin; Farsiah, Laina; Husaini, Husaini; Zulfan, Zulfan
CYBERSPACE: Jurnal Pendidikan Teknologi Informasi Vol 9 No 1 (2025)
Publisher : Universitas Islam Negeri Ar-Raniry Banda Aceh

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22373/cj.v9i1.28944

Abstract

Education plays a critical role in shaping career decisions for the future. However, many students encounter difficulties in selecting suitable academic programs, often stemming from a lack of confidence in their ability to make appropriate decisions. Consequently, students may choose study programs that do not align with their personal characteristics. This study emphasizes the importance of providing comprehensive information about various academic programs offered in higher education and developing tools to assist prospective students in making informed decisions. To address these challenges, a recommendation system using Hybrid Filtering technology has been developed. The system integrates Content-Based Filtering and Collaborative Filtering methods within the TensorFlow Recommenders System (TFRS) framework. The study utilized data from undergraduate students of the Faculty of Mathematics and Natural Sciences (FMIPA) across seven academic programs. By employing 10 features representing students' interests and talents, the recommendation system generated accurate and tailored suggestions for study programs. The model was trained and evaluated using both real and augmented (augmented) datasets with predefined hyperparameters. Results demonstrated that using only the real dataset achieved a Top-1 accuracy of 0.59 and a Top-5 accuracy of 0.97. When incorporating the augmented dataset, the Top-1 accuracy improved to 0.66, while the Top-5 accuracy reached 1.0. The findings reveal that combining real and augmented datasets enhances average accuracy by approximately 10% compared to using the real dataset alone. Additionally, the study program recommendations produced by the model showed significant improvement in quality. A web-based recommendation system utilizing the TFRS model was developed and positively evaluated by FMIPA students. User feedback indicated high satisfaction with the system's recommendations, demonstrating its effectiveness in guiding students toward suitable academic programs.