Garuda - Garba Rujukan Digital

Prediction of the Change Rate of Tumor Cells, Healthy Host Cells, and Effector Immune Cells in a Three-Dimensional Cancer Model using Extended Kalman Filter Fitriyati, Nina; Faizah, Salma Abidah; Sutanto, Taufik Edy
Jambura Journal of Biomathematics (JJBM) Volume 5, Issue 1: June 2024
Publisher : Department of Mathematics, Universitas Negeri Gorontalo

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37905/jjbm.v5i1.24672

In this study, we develop and implement the Extended Kalman Filter (EKF) to forecast the rate of change in tumor cells, healthy host cells, and effector immune cells within the Itik-Banks model. This novel application of EKF in cancer dynamics modeling aims to provide precise real-time estimations of cellular interactions, especially in constructing a new state space representation from the Itik-Banks model. We use a first-order Taylor series to linearize the model. The numerical simulations were performed to analyze the accuracy of this new state space with data from William Gilpin's GitHub repository. The results show that the EKF predictions strongly align with actual data, i.e., in the prior and posterior steps for tumor and healthy host cells, there is a strong agreement between the predictions and the actual data. The EKF captures the oscillatory nature of the tumor and healthy host cell population well. The peaks and troughs of the predictions align closely with the actual data, indicating the EKF's effectiveness in modeling the dynamic behavior of the tumor and healthy host cells. However, for effector immune cells, the oscillatory nature of the data in these cells gives rise to slight deviations. This represents a significant challenge in the future for updating the state space representations. Despite minor discrepancies, the EKF demonstrates a strong performance in both the training and testing data, with the posterior step estimates significantly improving the prior step accuracy. This study emphasizes the importance of data availability for accurate predictions, noting a symmetric Mean Absolute Percentage Error (sMAPE) of 35.92% when data is unavailable. Prompt correction with new data is essential to maintain accuracy. This research underscores the EKF's potential for real-time monitoring and prediction in complex biological systems.

Analysis of the Achievement of Program Learning Outcomes Based on an Outcome-Based Education Curriculum Manaqib, Muhammad; Sidqi, Serin Tias; Sutanto, Taufik Edy; Elfiyanti, Gustina; Mahmudi, Mahmudi
International Journal of Innovation and Education Research Vol. 3 No. 1 (2024)
Publisher : Universitas Bengkulu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33369/ijier.v3i1.34389

The curriculum, a set of plans and arrangements for learning outcomes, study materials, processes, and assessments, serves as a guide for implementing study programs. The Outcome Based Education (OBE) curriculum has been adopted at the higher education level to keep pace with rapid technological developments. Program Learning Outcomes (PLOs) are designed to articulate learning objectives into measurable and assessable statements using OBE principles. The results of PLO evaluations can be used to enhance PLO standards or quality performance and for accreditation purposes. Syarif Hidayatullah State Islamic University Jakarta has implemented the OBE curriculum in study programs in a significant stride towards international accreditation. This research, conducted using a quantitative research method, aims to analyze the achievement of the PLOs established by the Bachelor of Mathematics Study Program, Faculty of Science and Technology, Syarif Hidayatullah State Islamic University Jakarta. The research method involves measuring the achievement of PLOs by analyzing student grades and providing a comprehensive and objective assessment of the PLOs established by the Bachelor of Mathematics Program. The achievement of PLOs in the Bachelor of Mathematics Study Program, Faculty of Science and Technology, Syarif Hidayatullah State Islamic University Jakarta, is a significant 85.11%.

Active learning on Indonesian Twitter sentiment analysis using uncertainty sampling Liebenlito, Muhaza; Inayah, Nur; Choerunnisa, Esti; Sutanto, Taufik Edy; Inna, Suma
Journal of Applied Data Sciences Vol 5, No 1: JANUARY 2024
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v5i1.144

Nowadays, sentiment analysis research in social media is rapidly developing. Sentiment analysis typically falls under supervised learning, which requires annotating data. However, the annotation process for sentiment analysis tasks is notoriously time-consuming. Fortunately, an effective strategy to overcome this challenge has emerged, known as active learning. Active learning involves labeling only a small subset of the dataset, leaving the rest for annotation through sampling strategies. This study focuses on comparing two active learning strategies: random sampling and boundary sampling. These strategies are applied to machine learning models such as logistic regression and random forests. In addition, we present an evaluation of the model performance and data savings achieved by implementing these strategies in the context of traditional machine learning for sentiment analysis on Twitter. The dataset considered consists of two labels: positive and negative sentiments. The results of our investigation show that active learning can significantly reduce the amount of training data required, saving up to 65% of the total training data required to achieve peak model accuracy. The most successful model identified uses a random forest with a margin sampling strategy, yielding an accuracy of 81.12% and an F1 score of 88.60%. This research highlights the effectiveness of active learning strategies in sentiment analysis, demonstrating their potential to improve model performance and resource efficiency. The results underscore the viability of employing active learning methods, particularly the combination of random forest models with margin sampling, for more efficient sentiment analysis in social media.

Web Traffic Anomaly Detection using Stacked Long Short-Term Memory Rahman, Fathu; Sutanto, Taufik Edy; Fitriyati, Nina
InPrime: Indonesian Journal of Pure and Applied Mathematics Vol 3, No 2 (2021)
Publisher : Department of Mathematics, Faculty of Sciences and Technology, UIN Syarif Hidayatullah

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15408/inprime.v3i2.21879

AbstractAn example of anomaly detection is detecting behavioral deviations in internet use. This behavior can be seen from web traffic, which is the amount of data sent and received by people who visit websites. In this study, anomaly detection was carried out using stacked Long Short-Term Memory (LSTM). First, stacked LSTM is used to create forecasting models using training data. Then the error value generated from the prediction on test data is used to perform anomaly detection. We conduct hyperparameter optimization on sliding window parameter. Sliding window is a sub-sequential data of time-series data used as input in the prediction model. The case study was conducted on the real Yahoo Webscope S5 web traffic dataset, consisting of 67 datasets, each of which has three features, namely timestamp, value, and anomaly label. The result shows that the average sensitivity is 0.834 and the average Area Under ROC Curve (AUC) is 0.931. In addition, for some of the data used, the window size selection can affect the sum of the sensitivity and AUC values. In this study, anomaly detection using stacked LSTM is described in detail and can be used for anomaly detection in other similar problems.Keywords: time-series data; sliding window; web traffic; window size. AbstrakSalah satu contoh deteksi anomali adalah mendeteksi penyimpangan perilaku dalam penggunaan internet. Perilaku ini dapat dilihat dari web traffic, yaitu jumlah data yang dikirim dan diterima oleh orang-orang yang mengunjungi situs web. Pada penelitian ini, deteksi anomali dilakukan menggunakan Long Short-Term Mermory (LSTM) bertumpuk. Pertama, LSTM bertumpuk digunakan untuk membuat model peramalan menggunakan data latih. Kemudian nilai error yang dihasilkan dari prediksi pada data uji digunakan untuk melakukan deteksi anomali. Kami melakukan optimasi hyperparameter pada parameter sliding window. Sliding window adalah data sub-sekuensial dari data runtun waktu yang digunakan sebagai input pada model prediksi. Studi kasus dilakukan pada dataset web traffic Yahoo Webscope S5 yang terdiri dari 67 dataset yang masing-masing memiliki tiga fitur yaitu timestamp, value, dan anomaly label. Hasil menunjukkan bahwa rata-rata sensitivitas sebesar 0.834 dan rata-rata Area Under ROC Curve (AUC) sebesar 0.931. Selain itu, untuk beberapa data yang digunakan, pemilihan window size dapat mempengaruhi jumlah dari nilai sensitivitas dan AUC. Pada penelitian ini, deteksi anomali menggunakan LSTM bertumpuk dijelaskan secara rinci dan dapat digunakan untuk deteksi anomali pada permasalahan lainnya yang serupa.Kata kunci: data runtun waktu; sliding window; web traffic; window size.

Exploratory Data Analysis of Indonesian Presidential Election Candidate Campaign in 2019 on Twitter Hermawan, Fadlan Bima; Sutanto, Taufik Edy; Santoso, Ary
Indonesian Journal of Artificial Intelligence and Data Mining Vol 7, No 2 (2024): September 2024
Publisher : Universitas Islam Negeri Sultan Syarif Kasim Riau

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24014/ijaidm.v7i2.26308

Social media users in Indonesia are growing over time, this has caused many political actors, both individuals, and political parties, to take advantage of this. According to Hootsuite (We are Social) in the Digital 2022 report, Indonesia has 191 million active social media users, so there will be many political actors campaigning on social media. In the literature that discusses similar topics, it is rare to analyze comprehensive exploratory data analysis such as text analysis, hashtags, and social network analysis. From the data analysis conducted in the 2019 Election, the following results were obtained. In text analysis, the narratives used by the two candidates were very different, it was seen that some used other positive and negative narratives. The hashtag analysis found that consistency in campaigning had an influence on electability in the election. In the analysis of social networks, it is found that users who influence social media campaigns, users who appear are not only politicians but also ordinary people who try to support their chosen candidate. The research conducted is expected to be a basic reference in exploring data on social media in other cases, especially in the political field

Studi Empiris Model BERT dan DistilBERT Analisis Sentimen pada Pemilihan Presiden Indonesia Mahira Putri; Sutanto, Taufik Edy; Inna, Suma
The Indonesian Journal of Computer Science Vol. 12 No. 5 (2023): The Indonesian Journal of Computer Science (IJCS)
Publisher : AI Society & STMIK Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33022/ijcs.v12i5.3445

Peningkatan jumlah pengguna media sosial di Indonesia sejak tahun 2014 menyebabkan data yang dihasilkan semakin besar dan kompleks, sehingga komputasi yang diperlukan untuk mengolahnya juga semakin besar. Untuk melakukan komputasi pada data yang besar diperlukan model yang kompatibel, efektif, dan efisien. Penelitian ini adalah kajian numerik dari dua model terbaik Deep Learning saat paper ini ditulis, yaitu BERT dan DistilBERT pada kasus analisis sentimen menggunakan ratusan ribu tweet terkait pemilihan presiden Indonesia tahun 2014 dan 2019. Analisis yang dilakukan meliputi waktu eksekusi dan konsumsi memori. Pada model dengan nilai hyperparameter optimal, tercatat bahwa DistilBERT melakukan proses pelatihan dan prediksi 84% lebih cepat dengan penggunaan memori GPU 79% lebih efisien dengan nilai akurasi tidak terpaut jauh, yaitu 0.89 dan 0.85 untuk BERT dan DistilBERT. Hasil kajian ini dapat digunakan untuk memperkirakan besarnya sumberdaya komputasi atau biaya yang dibutuhkan ketika menggunakan model BERT atau DistilBERT pada data yang besar.

Analisis Tweet Politik-Keagamaan pada Hasil Pemilihan Presiden Indonesia tahun 2014 dan 2019: Sebuah Studi Eksploratif Zain, Poppy Dalama; Sutanto, Taufik Edy; Liebenlito, Muhaza
The Indonesian Journal of Computer Science Vol. 13 No. 1 (2024): The Indonesian Journal of Computer Science (IJCS)
Publisher : AI Society & STMIK Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33022/ijcs.v13i1.3751

Agama dan politik saling terkait dalam konteks pemilihan umum di Indonesia. Penggunaan isu agama dalam kegiatan politik merambah luas di pemilu Indonesia tahun 2014 dan 2019 di media sosial. Paper ini mengisi kekosongan di literatur untuk mengkaji fenomena penggunaan isu agama dalam politik secara kuantitatif. Data diambil menggunakan dari media sosial twitter menggunakan API secara legal dan menjaga privasi pengguna. Pengambilan data dilakukan dengan menggunakan kata kunci terkait agama seperti Islam, Al-Qur’an, Hadits, Halal, Shalat, dan sebagainya lalu di filter dengan berbagai kata kunci terkait politik. Melalui berbagai teknik eksplorasi data seperti analisis korelasi Spearman dan visualisasi geospasial, penelitian ini menemukan adanya hubungan signifikan antara banyaknya isu agama terkait politik dan perolehan suara calon presiden. Pada tahun 2014 korelasi untuk Prabowo dibandingkan korelasi untuk Jokowi lebih tinggi yaitu sebesar 0.72, lalu menurun pada tahun 2019 menjadi 0.56. Penelitian ini dapat dijadikan inspirasi untuk penurunan dan pencegahan terjadinya polarisasi di masyarakat akibat penggunaan isu agama dalam kegiatan politik.

Web Traffic Anomaly Detection using Stacked Long Short-Term Memory Rahman, Fathu; Sutanto, Taufik Edy; Fitriyati, Nina
InPrime: Indonesian Journal of Pure and Applied Mathematics Vol. 3 No. 2 (2021)
Publisher : Department of Mathematics, Faculty of Sciences and Technology, UIN Syarif Hidayatullah

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15408/inprime.v3i2.21879

AbstractAn example of anomaly detection is detecting behavioral deviations in internet use. This behavior can be seen from web traffic, which is the amount of data sent and received by people who visit websites. In this study, anomaly detection was carried out using stacked Long Short-Term Memory (LSTM). First, stacked LSTM is used to create forecasting models using training data. Then the error value generated from the prediction on test data is used to perform anomaly detection. We conduct hyperparameter optimization on sliding window parameter. Sliding window is a sub-sequential data of time-series data used as input in the prediction model. The case study was conducted on the real Yahoo Webscope S5 web traffic dataset, consisting of 67 datasets, each of which has three features, namely timestamp, value, and anomaly label. The result shows that the average sensitivity is 0.834 and the average Area Under ROC Curve (AUC) is 0.931. In addition, for some of the data used, the window size selection can affect the sum of the sensitivity and AUC values. In this study, anomaly detection using stacked LSTM is described in detail and can be used for anomaly detection in other similar problems.Keywords: time-series data; sliding window; web traffic; window size. AbstrakSalah satu contoh deteksi anomali adalah mendeteksi penyimpangan perilaku dalam penggunaan internet. Perilaku ini dapat dilihat dari web traffic, yaitu jumlah data yang dikirim dan diterima oleh orang-orang yang mengunjungi situs web. Pada penelitian ini, deteksi anomali dilakukan menggunakan Long Short-Term Mermory (LSTM) bertumpuk. Pertama, LSTM bertumpuk digunakan untuk membuat model peramalan menggunakan data latih. Kemudian nilai error yang dihasilkan dari prediksi pada data uji digunakan untuk melakukan deteksi anomali. Kami melakukan optimasi hyperparameter pada parameter sliding window. Sliding window adalah data sub-sekuensial dari data runtun waktu yang digunakan sebagai input pada model prediksi. Studi kasus dilakukan pada dataset web traffic Yahoo Webscope S5 yang terdiri dari 67 dataset yang masing-masing memiliki tiga fitur yaitu timestamp, value, dan anomaly label. Hasil menunjukkan bahwa rata-rata sensitivitas sebesar 0.834 dan rata-rata Area Under ROC Curve (AUC) sebesar 0.931. Selain itu, untuk beberapa data yang digunakan, pemilihan window size dapat mempengaruhi jumlah dari nilai sensitivitas dan AUC. Pada penelitian ini, deteksi anomali menggunakan LSTM bertumpuk dijelaskan secara rinci dan dapat digunakan untuk deteksi anomali pada permasalahan lainnya yang serupa.Kata kunci: data runtun waktu; sliding window; web traffic; window size.

Analisis Performa Klaster Single Board Computer dalam Implementasi Singular Value Decomposition Azka, Syahrul; Liebenlito, Muhaza; Sutanto, Taufik Edy
Euler : Jurnal Ilmiah Matematika, Sains dan Teknologi Volume 13 Issue 3 December 2025
Publisher : Universitas Negeri Gorontalo

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37905/euler.v13i3.33367

This study aims to evaluate the performance of Singular Value Decomposition operations based on the divide-and-conquer method on two computing cluster architectures: an Intel Core i5-12400-based PC cluster and an Allwinner H618-based Orange Pi Zero 3 Single Board Computer cluster. The evaluation focuses on three key metrics: execution time, speedup, and energy consumption. Experiments were conducted on three matrix sizes (2160×2160, 3240×3240, and 5400×5400) with processor cores ranging from 1 to 12. Energy consumption was measured using a wattmeter by recording peak power during execution. The results show that the PC cluster achieves faster execution times but exhibits limited parallel scalability, reaching a maximum speedup of 10.31× and energy consumption of 2.07 Wh for the 5400×5400 matrix with 12 cores. In contrast, the SBC cluster demonstrates significantly higher parallel efficiency, achieving a speedup of 117.75× with energy consumption of only 0.23 Wh under the same configuration. These findings indicate that the SBC cluster offers a promising energy-efficient, cost-effective solution for parallel numerical computing, particularly for sustainable computing infrastructure in higher education, in alignment with the Sustainable Development Goals 7.

Title

Found 9 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 9 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 9 Documents
Search