Claim Missing Document
Check
Articles

Found 38 Documents
Search

Analisis Komparatif Algoritma LSTM, GRU, BiGRU, dan BiLSTM Untuk Prediksi Degradasi Bearing Turbin PLTU Raymond, Rifky; Saputra, Neva; Tupamahu, Meldrin; Herawati, Neng Ayu; Purwarianti, Ayu; Utama, Nugraha Priya
Jurnal Infomedia: Teknik Informatika, Multimedia, dan Jaringan Vol 10, No 1 (2025): Jurnal Infomedia
Publisher : Politeknik Negeri Lhokseumawe

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30811/jim.v10i1.7127

Abstract

Pembangkit Listrik Tenaga Uap (PLTU) merupakan salah satu sumber utama pasokan listrik nasional, di mana keandalan komponen kritis seperti bearing turbin sangat menentukan kontinuitas operasional. Kegagalan pada bearing dapat menyebabkan downtime tidak terduga dan kerugian biaya yang signifikan. Oleh karena itu, pendekatan predictive maintenance menjadi strategi penting dalam memitigasi potensi kegagalan tersebut. Penelitian ini bertujuan untuk membandingkan performa empat algoritma deep learning yaitu LSTM, GRU, BiGRU, dan BiLSTM dalam memprediksi Remaining Useful Life (RUL) dari bearing turbin. Data yang digunakan merupakan data sensor aktual dari pembangkit, yang telah direduksi dimensinya menggunakan Principal Component Analysis (PCA) untuk membentuk Health Index sebagai representasi degradasi peralatan. Evaluasi dilakukan menggunakan metrik MAE (Mean Absolute Error) dan RMSE (Root Mean Squared Error). Hasil eksperimen menunjukkan bahwa model BiLSTM memiliki performa terbaik dengan nilai MAE sebesar 0.27 dan RMSE sebesar 0.37. Penelitian ini berkontribusi dalam menyediakan panduan pemilihan model prediksi RUL berbasis data sensor riil pada peralatan PLTU, yang mendukung penerapan pemeliharaan prediktif secara lebih akurat dan efisien
Causal Discovery of ICU Stay Length: PC Algorithm Approach with ICD-Lab Data Halim, Ismail Syababun; Martalia, Anastasia Mia; Hibatullah, Muhammad Helmi; Utama, Nugraha Priya; Purwarianti, Ayu
JTERA (Jurnal Teknologi Rekayasa) Vol 10, No 1: Juni 2025
Publisher : Politeknik Sukabumi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31544/jtera.v10.i1.2025.173-182

Abstract

Dalam sistem pelayanan kesehatan, Unit Perawatan Intensif (ICU) merupakan komponen penting untuk menangani pasien dalam kondisi kritis yang membutuhkan pemantauan intensif. Namun, durasi rawat inap atau Length of Stay (LoS) seorang pasien di ICU biasanya sangat bervariasi dan perpanjangan LoS berdampak signifikan pada beban biaya, penggunaan sumber daya, dan efisiensi pelayanan di rumah sakit. Penelitian ini bertujuan untuk mengidentifikasi faktor kausal yang memengaruhi LoS ICU menggunakan Algoritma Peter-Clark (PC) untuk penemuan kausal atau Causal Discovery. Data yang digunakan berasal dari MIMIC-IV, sebuah basis data klinis komprehensif dari Beth Israel Deaconess Medical Center tahun 2008–2019, yang mencakup demografi pasien, kode diagnosis ICD, dan hasil pemeriksaan laboratorium. Metode yang digunakan meliputi penerapan Algoritma PC, yang dipilih karena kemampuannya pada data berdimensi tinggi dengan Fisher's Z-test untuk pengujian independensi, yang diimplementasikan pada berbagai tingkat signifikansi (ɑ = 0.01, 0.05, 0.1). Validasi dilakukan melalui 500 iterasi bootstrap untuk mengetahui stabilitas dari struktur graf kausal. Hasil analisis menunjukkan enam variabel yang secara konsisten menjadi penyebab langsung LoS ICU diantaranya admission type, APR-DRG severity, high mortality risk, category, flag, dan anchor age. Struktur kausal yang dihasilkan memberikan gambaran hubungan sebab-akibat yang stabil dan signifikan antar variabel klinis, yang dapat digunakan untuk mendukung pengambilan keputusan berbasis data dalam manajemen pasien dan alokasi sumber daya ICU. Studi ini juga menegaskan potensi pendekatan Causal Discovery dalam analitik layanan kesehatan, khususnya dalam memahami faktor determinan LoS ICU secara mendalam.
Exploring The Effectiveness of In-Context Methods in Human-Aligned Large Language Models Across Languages Prathama, Ubaidillah Ariq; Ayu Purwarianti; Samuel Cahyawijaya
JUTI: Jurnal Ilmiah Teknologi Informasi Vol.23, No.2, July 2025
Publisher : Department of Informatics, Institut Teknologi Sepuluh Nopember

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j24068535.v23i2.a1323

Abstract

Most of past studies about in-context methods like in-context learning (ICL), cross-lingual ICL (X-ICL), and in-context alignment (ICA) come from older, unaligned large language models (LLMs). However, modern human-aligned LLMs are different; they come with chat-style prompt templates, are extensively human-aligned, and cover many more languages. We re-examined these in-context techniques using two recent, human-aligned multilingual LLMs. Our study covered 20 languages from seven different language families, representing high, mid, and low-resource levels. We tested how well these methods generalized using two tasks: topic classification (SIB-200) and machine reading comprehension (Belebele). We found that utilizing prompt templates significantly improves the performance of both ICL and X-ICL. Furthermore, ICA proves particularly effective for mid- and low-resource languages, boosting their f1-score by up to 6.1%. For X-ICL, choosing a source language that is linguistically similar to the target language, rather than defaulting to English, can lead to substantial gains, with improvements reaching up to 21.98%. Semantically similar ICL examples continue to be highly relevant for human-aligned LLMs, providing up to a 31.42% advantage over static examples. However, this gain decreases when using machine translation model to translate query from target language. These results collectively suggest that while modern human-aligned LLMs definitely benefit from in-context information, the extent of these gains is highly dependent on careful prompt design, the language's resource level, language pairing, and the overall complexity of the task.
EKSTRAKSI KATA KUNCI OTOMATIS UNTUK DOKUMEN BAHASA INDONESIA STUDI KASUS: ARTIKEL JURNAL ILMIAH KOLEKSI PDII LIPI Sari, Diana; Purwarianti, Ayu
BACA: Jurnal Dokumentasi dan Informasi Vol. 35 No. 2 (2014): BACA: Jurnal Dokumentasi dan Informasi (Desember)
Publisher : Direktorat Repositori, Multimedia, dan Penerbitan Ilmiah - Badan Riset dan Inovasi Nasional (BRIN Publishing)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.14203/j.baca.v35i2.192

Abstract

Keyword determination by using controlled vocabulary is not a difficult task for information analysts. However,specify keywords for hundreds or even thousands of articles will take time and effort of the analysts. To ease thework, it needs to be made a system of automatic keyword extraction. The construction of this system passes thestages of preprocessing, translating, and pinpointing keyword candidates with a list of keywords. The research wascarried out by using 33 articles taken from PDII LIPI journal collections. This research employed 3 weighing methods,namely TF, TF x IDF and WIDF. The best result was obtained from TF x IDF method. To refine the result, the authorcarried out fixing the keywords results and using levensthein algorithm.
Retrieval-Augmented Generation (RAG) Chatbot for Handling Customer Complaints in the Energy Sector Haryono Putro, Ibnu Prastowo; Antoni, Jefry; Adhitya, Maulana Krisna; Herawati, Neng Ayu; Purwarianti, Ayu; Utama, Nugraha Priya
Jurnal Infomedia: Teknik Informatika, Multimedia, dan Jaringan Vol 10, No 2 (2025): Jurnal Infomedia
Publisher : Politeknik Negeri Lhokseumawe

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30811/jim.v10i2.7169

Abstract

Fast and accurate customer service is critical in the energy sector, especially for large-scale utilities like PLN. This study introduces a novel Retrieval-Augmented Generation (RAG)-based chatbot tailored for PLN’s internal operational context to automate customer complaint resolution in Bahasa Indonesia. In contrast to previous approaches that utilize only fine-tuned LLMs or retrieval-based question answering, our system uniquely integrates internal complaint records stored in internal database with a local Indonesian-optimized LLM through LangChain orchestration. The proposed architecture features temporal and linguistic preprocessing, vector embedding using FAISS, and a dynamic clarification-fallback mechanism, ensuring context-aware and grounded responses. This work contributes a scalable framework for deploying generative AI in high-stakes public utility settings, emphasizing data privacy, language fidelity, and real-time applicability. Evaluation results both simulated and human-reviewed demonstrate the chatbot’s effectiveness, achieving BLEU-4 of 46.5 and ROUGE-L of 0.63, with 92% of answers rated helpful. These findings underscore the model's potential to enhance customer experience and operational efficiency in Indonesia’s energy sector.
Analysis of Defense Mechanisms Against FGSM Adversarial Attacks on ResNet Deep Learning Models Using the CIFAR-10 Dataset Miranti Jatnika Riski; Krishna Aurelio Noviandri; Yoga Hanggara; Nugraha Priya Utama; Ayu Purwarianti
Jurnal Sistem Cerdas Vol. 8 No. 2 (2025): August
Publisher : APIC

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37396/jsc.v8i2.527

Abstract

Adversarial attacks threaten the reliability of deep learning models in image classification, requiring effective defense mechanisms. This study evaluates how defense distillation and adversarial training protect ResNet18 models trained on CIFAR-10 data against Fast Gradient Sign Method (FGSM) attacks. The baseline model achieves 85.01% accuracy on clean data but its accuracy falls to 19.23% when FGSM attacks at epsilon 0.3. The accuracy of defense distillation drops to 23.68% when epsilon reaches 0.3 but adversarial training maintains 0.34% accuracy at epsilon 0.25 although it reduces clean data accuracy to 57.08%. The analysis shows that classes with similar visual characteristics such as cats and dogs remain vulnerable to attacks. The study demonstrates the requirement for balanced defense approaches while indicating additional work needs to improve model robustness.
Detecting Type and Index Mutation in Cancer DNA Sequence Based on Needleman–Wunsch Algorithm Wisesty, Untari Novia; Mengko, Tati Rajab; Purwarianti, Ayu; Pancoro, Adi
Jurnal Ilmu Komputer dan Informasi Vol. 17 No. 2 (2024): Jurnal Ilmu Komputer dan Informasi (Journal of Computer Science and Informatio
Publisher : Faculty of Computer Science - Universitas Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21609/jiki.v17i2.1273

Abstract

Detecting DNA sequence mutations in cancer patients contributes to early identification and treatment of the disease, which ultimately enhances the effectiveness of treatment. Bioinformatics utilizes sequence alignment as a powerful tool for identifying mutations in DNA sequences. We used the Needleman-Wunsch algorithm to identify mutations in DNA sequence data from cancer patients. The cancer sequence dataset used includes breast, cervix uteri, lung, colon, liver and prostate cancer. Various types of mutations were identified, such as Single Nucleotide Variant (SNV)/substitution, insertion, and deletion, locate by the nucleotide index. The Needleman Wunch algorithm can detect type and index mutation with the average F1-scores 0.9507 for all types of mutations, 0.9919 for SNV, 0.7554 for insertion, and 0.8658 for deletion with a tolerance of 5 bp. The F1-scores obtained are not correlated with gene length. The time required ranges from 1.03 seconds for a 290 base pair gene to 3211.45 seconds for a gene with 16613 base pairs.
Enhancing Natural Language Inference Performance with Knowledge Graph for COVID-19 Automated Fact-Checking in Indonesian Language Muharram, Arief Purnama; Purwarianti, Ayu
Journal of ICT Research and Applications Vol. 19 No. 1 (2025)
Publisher : DRPM - ITB

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.5614/itbj.ict.res.appl.2025.19.1.2

Abstract

Automated fact-checking is a key strategy to overcome the spread of COVID-19 misinformation on the internet. These systems typically leverage deep learning approaches through natural language inference (NLI) to verify the truthfulness of information based on supporting evidence. However, one challenge that arises in deep learning is performance stagnation due to a lack of knowledge during training. This study proposes using a knowledge graph (KG) as external knowledge to enhance NLI performance for automated COVID-19 fact-checking in the Indonesian language. The proposed model architecture comprises three modules: a fact module, an NLI module, and a classifier module. The fact module processes information from the KG, while the NLI module handles semantic relationships between the given premise and hypothesis. The representation vectors from both modules are concatenated and fed into the classifier module to produce the final result. The model was trained using the generated Indonesian COVID-19 fact-checking dataset and the COVID-19 KG Bahasa Indonesia. Our study demonstrates that incorporating KGs can significantly improve NLI performance in fact-checking, achieving a maximum accuracy of 0.8616. This suggests that KGs are a valuable component for enhancing NLI performance in automated fact-checking.
Co-Authors Adhitia, Rama Adhitya, Maulana Krisna Adi Pancoro Afrida Helen Alit Kesatria Mendala Aminuddin, Amir Andri Fachrur Rozie Andria Arisal Antoni, Jefry Arief Rahman Arry Akhmad Arman Arry Akhmad Arman Aswin Juari Dessi Puji Lestari Devi Munandar Diana Permata Sari Diana Sari Dianadewi Riswantini Dwi H. Widyantoro Dwi Hendratmo Widyantoro Ekasari Nugraheni Hadi Saputra Halim, Ismail Syababun Hari Bagus Firdaud Haryono Putro, Ibnu Prastowo Hendradjaja, Bayu Herawati, Neng Ayu Herry Sujaini Herry Sujaini Hibatullah, Muhammad Helmi Imamah Imamah Intan Jelita Saragih Iping S Suwardi Iping Supriana Iping Supriana Iping Supriana Iping Supriana Iping Supriana Iping Supriana Iping Supriana Irfan Afif Iskandar, Yanti Rubiyanti Kridanto Surendro Krishna Aurelio Noviandri KUSPRIYANTO Lisa Madlberger Martalia, Anastasia Mia Mengko, Tati Rajab Miranti Jatnika Riski Mochammad Ibrahim Muh Habibi Haidir Muharram, Arief Purnama Neng Ayu Herawati Nila Nurmala Novi Yusliani Nugraha Priya Utama Prathama, Ubaidillah Ariq Purnomo Husnul Khotimah Putra, Helmy Satria Martha Rama Adhitia Rama Adhitia Rama Adhitia Rifky Raymond, Rifky Ririn Dwi Agustin Ruskanda, Fariska Zakhralativa Samuel Cahyawijaya Saputra, Neva Supanto Supanto Titin Pramiyati Titin Pramiyati Titin Pramiyati Titin Pramiyati Tupamahu, Meldrin Untari Novia Wisesty Utama, Nugraha Priya Wiwin Suwarningsih Wiwin Suwarningsih wiwin suwarningsih Wiwin Suwarningsih Wiwin Suwarningsih Yoga Hanggara Yuhana, Umi Laili Zakiy Firdaus Alfikri Zakiy Firdaus Alfikri