Garuda - Garba Rujukan Digital

An Adaptive Stacking An Adaptive Stacking Approach for Monthly Rainfall Prediction with Hybrid Feature Selection: Hybrid Feature Selection Zulfa, Ahmad; Saikhu, Ahmad; Pradana, Hilmil; Budiawan, Irvan
ULTIMA Computing Vol 17 No 1 (2025): Ultima Computing: Jurnal Sistem Komputer
Publisher : Faculty of Engineering and Informatics, Universitas Multimedia Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31937/sk.v17i1.4157

Rainfall is a critical climatic element for water resource management, agriculture, and hydrometeorological disaster mitigation. However, its nonlinear and fluctuating characteristics require a careful and adaptive predictive approach. This study aims to develop a monthly rainfall prediction model using an Adaptive Stacking Ensemble method combined with a hybrid feature selection framework. The feature selection integrates three techniques”correlation analysis, feature importance from Random Forest, and Recursive Feature Elimination (RFE)”through a voting mechanism. Three machine learning algorithms, namely Random Forest, K-Nearest Neighbors (KNN), and XGBoost, are used as base learners. The meta-learner is adaptively selected based on the best-performing base model. Model performance is evaluated using R², RMSE, and MAE metrics. The proposed method is expected to produce a more accurate, stable, and reliable predictive model to support climate-based decision-making. By leveraging the hybrid feature selection framework, the model effectively identifies the most relevant weather variables related to monthly rainfall patterns, thereby reducing model complexity without sacrificing accuracy. The adaptive stacking approach also offers flexibility in capturing nonlinear relationships between features and targets, while enhancing model generalization across seasonally varying data. Experiments were conducted on long-term weather datasets, and the results demonstrate that the proposed model outperforms single models and conventional ensemble methods. This research contributes to the development of more robust, data-driven climate prediction systems that can be applied across sectors affected by rainfall variability.

PEMODELAN PERUBAHAN LAHAN DAN TUTUPAN LAHAN BERBASIS MARKOV-CHAIN DI KABUPATEN GUNUNG KIDUL Rahmawati, Septi Sri; Setyowati, Ratih; Ramlah; Azizah, Salsabila Nur; Ardiansyah, Ramadhani Muhammad Yusuf; Saikhu, Ahmad
Jurnal Tanah dan Sumberdaya Lahan Vol. 12 No. 2 (2025)
Publisher : Departemen Tanah, Fakultas Pertanian, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21776/ub.jtsl.2025.012.2.19

The standard values of quality and quantity of living needs continue to increase over time, leading to competition in the utilization of agricultural and non-agricultural land. This results in uncontrolled land use conversion. Land use changes in Gunungkidul between the 1940s and 1970s caused widespread deforestation, making the area barren. Conservation efforts up to the early 2000s yielded positive results. However, population growth and the expansion of activity centers in Gunungkidul during the 2000s may trigger further deforestation. This study analyzes spatiotemporal land use changes in Gunungkidul Regency over the period 2015–2023. Land use data were obtained from Landsat and Sentinel-2 satellite imagery and analyzed using Geographic Information Systems (GIS) and accuracy assessment through ground checks. The results show significant land use changes, particularly the conversion of vegetation into built-up areas, with an increase/decrease in area from 2015 to 2023 of 72.65%. The most significant changes occurred in Wates District, the administrative center of Gunungkidul Regency, forming a pattern concentrated around service centers and spreading linearly along access routes to these centers. Land changes were also observed in the northern and southern regions with a scattered pattern. These changes are driven by population growth and regional development in Gunungkidul Regency. Land use changes may lead to land degradation, highlighting the importance of this study in providing crucial information for better spatial planning and land management in the future.

THE EFFECT OF FACIAL ACCESSORY AUGMENTATION ON THE ACCURACY OF DEEP LEARNING-BASED FACIAL RECOGNITION SYSTEMS Hidayat, Ahmad Nur; Suciati, Nanik; Saikhu, Ahmad
JURTEKSI (Jurnal Teknologi dan Sistem Informasi) Vol 11, No 3 (2025): Juni 2025
Publisher : Universitas Royal

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33330/jurteksi.v11i3.3846

Abstract: Face recognition based on deep learning has become an important technology in many areas. However, these systems often face challenges in real-world conditions, such as when the face is partially covered by accessories such as masks or glasses. This study aims to evaluate the effect of data augmentation by adding facial accessories (masks, glasses, and a combination of both) and geometric augmentation on the accuracy of face recognition systems. There are three types of datasets used in this method: the original dataset (category 1), the dataset with facial accessories augmentation (category 2), and the dataset with geometric augmentation (category 3). Data augmentation was performed on the training dataset to increase diversity, followed by the face detection process using SCRFD and feature extraction with ArcFace. The model was then trained using Multi-Layer Perceptron (MLP). Based on the results, adding face accessories (category 2) made the model a lot more accurate, hitting 99% accuracy. In category 3, adding geometric features improved accuracy to 91%. Other evaluation metrics, such as precision, recall, and F1-score, also showed improvement after augmentation. This study concludes that facial accessories augmentation is more effective in improving the accuracy and robustness of face recognition models compared to geometric augmentation.Keywords: augmentation; deep learning; face recognition; glasses. Abstrak: Pengenalan wajah berbasis deep learning telah menjadi salah satu teknologi penting dalam berbagai aplikasi. Namun, sistem ini sering kali menghadapi tantangan dalam kondisi dunia nyata, seperti saat wajah tertutup sebagian oleh aksesori seperti masker atau kacamata. Penelitian ini bertujuan untuk mengevaluasi pengaruh augmentasi data dengan menambahkan aksesori wajah (masker, kacamata, dan kombinasi keduanya) serta augmentasi geometris terhadap akurasi sistem pengenalan wajah. Metode yang digunakan melibatkan tiga kategori dataset: dataset asli tanpa augmentasi (kategori 1), dataset dengan augmentasi aksesoris wajah (kategori 2), dan dataset dengan augmentasi geometris (kategori 3). Augmentasi data dilakukan pada dataset pelatihan untuk meningkatkan keberagaman, diikuti dengan proses deteksi wajah menggunakan SCRFD dan ekstraksi fitur dengan ArcFace. Model kemudian dilatih menggunakan Multi-Layer Perceptron (MLP). Hasil penelitian menunjukkan bahwa augmentasi aksesoris wajah (kategori 2) memberikan peningkatan signifikan pada akurasi model, mencapai 99%, sedangkan kategori 3 dengan augmentasi geometris mencapai akurasi 91%. Metrik evaluasi lainnya, seperti precision, recall, dan F1-score, juga menunjukkan peningkatan setelah augmentasi. Penelitian ini menyimpulkan bahwa augmentasi aksesoris wajah lebih efektif dalam meningkatkan akurasi dan ketahanan model pengenalan wajah dibandingkan dengan augmentasi geometris.Kata kunci: augmentasi; deep learning; kacamata; pengenalan wajah.

PEMODELAN DATA RADIOSONDE MENGGUNAKAN STACKING ENSEMBLE UNTUK KLASIFIKASI HUJAN Hermansyah, Muhammad; Saikhu, Ahmad; Amaliah, Bilqis
JIPI (Jurnal Ilmiah Penelitian dan Pembelajaran Informatika) Vol 10, No 2 (2025)
Publisher : STKIP PGRI Tulungagung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29100/jipi.v10i2.7806

Perubahan iklim telah meningkatkan frekuensi dan intensitas kejadian cuaca ekstrem di wilayah tropis seperti Indonesia, sehingga men-imbulkan tantangan dalam pemanfaatan data observasi meteorologi untuk mitigasi bencana hidrometeorologis. Data observasi permukaan sering kali kurang mampu merepresentasikan dinamika vertikal at-mosfer dalam analisis kejadian cuaca ekstrem, seperti hujan sedang hingga lebat. Penelitian ini bertujuan mengembangkan model klasifikasi intensitas hujan berbasis data observasi udara atas dari radiosonde dengan pendekatan stacking ensemble, yang mengintegrasikan algorit-ma Random Forest, XGBoost, LightGBM, dan SVM, serta menggunakan HistGradientBoosting sebagai meta-learner. Untuk mengatasi ketidakseimbangan kelas antara kondisi berawan-hujan ringan dan hujan sedang-lebat, diterapkan teknik Synthetic Minority Over-sampling Technique (SMOTE). Evaluasi performa dilakukan menggunakan metrik precision, recall, F1-score, dan kurva precision-recall. Hasil menunjukkan bahwa model stacking ensemble memberikan performa terbaik dengan nilai precision sebesar 0,9084, F1-score 0,8718, dan average precision untuk kelas hujan sedang-lebat sebesar 0,949, melampaui seluruh model individual. Temuan ini menegaskan keunggulan integrasi data atmosfer vertikal dan pendekatan multi-algorithm machine learning dalam mendeteksi hujan intensitas sedang hingga lebat secara lebih akurat. Model ini memiliki potensi tinggi untuk diimplementasikan dalam sistem peringatan dini cuaca ekstrem, khususnya di wilayah tropis dengan keterbatasan data observasi permukaan.

Enhancing Electricity Consumption Prediction with Deep Learning through Advanced Data Splitting Techniques Pratiwi, Adinda Putri; Ginardi, Raden Venantius Hari; Saikhu, Ahmad
International Journal of Artificial Intelligence Research Vol 8, No 2 (2024): December 2024
Publisher : Universitas Dharma Wacana

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29099/ijair.v8i2.1204

Energy consumption is increasing due to population growth and industrial activity, making electricity essential in human life. With limited natural resources, effective management of electrical resources is crucial to reduce energy usage amidst rising demand. The current trends on using deep learning as prediction can enhance the performances. To have good performance it needs correct preprocessing data, so it will produce a model with less overfitting. This research proposes a model using time-series cross-validation as the splitting data and correlation to choose the best features set for the prediction of electricity consumption. Experiments will compare time-series cross-validation and holdout methods to see the performances of splitting data and enhancing the multi-horizon data. The experiment used 8 sets of feature lists, which are paired in combination based on correlation to ensure the best features that are related. The result is splitting data using time-series cross-validation can maintain good perfomances on mode and holdout can maintain a good evaluation performance across the horizon. Feature sets that include temporal features have excellent results, especially when combined with features that have the strongest correlation relationship with electricity consumption, leading to an enhanced R2. Among all the models tested, CNN-GRU had the best model for multistep prediction across various every horizons and feature sets.

Hybrid Decomposition ICEEMDAN-EWT Deep Learning Framework for Wind Speed Forecasting Alif Hidayat, Dedi Arman; Aditya Pradana , Muhamad Hilmil Muchtar; Saikhu, Ahmad
Journal of Applied Informatics and Computing Vol. 9 No. 4 (2025): August 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i4.10241

Accurate wind speed forecasting plays a crucial role in supporting early warning systems for extreme wind events. However, the inherent non-linearity and non-stationarity of wind speed data pose significant challenges. This study addresses these issues by evaluating the effectiveness of targeted Empirical Wavelet Transform (EWT) denoising applied to specific Intrinsic Mode Functions (IMFs) derived from Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN). Daily wind speed data from 2000 to 2023 were decomposed using ICEEMDAN, and denoising was selectively applied to IMF1, IMF2, and IMF3. Each IMF was then modeled using a Bidirectional Long Short-Term Memory (BiLSTM) network under a time-series cross-validation framework. Among all model configurations, the ICEEMDAN+EWT(IMF1 & IMF2)+BiLSTM model achieved the highest predictive accuracy, with an R² of 0.8885, RMSE of 0.501, and MAPE of 7.64%. This result outperformed both the baseline BiLSTM model (R² = 0.0501) and the ICEEMDAN+BiLSTM model without EWT denoising (R² = 0.6433). Moreover, denoising on IMF1 alone also yielded a strong performance (R² = 0.8879), emphasizing the importance of early component selection. Conversely, applying EWT to IMF2 or IMF3 individually resulted in lower R² values of 0.6639 and 0.6327, respectively, indicating limited individual contribution. These findings confirm that selective denoising, especially on the high-frequency IMFs, substantially enhances forecasting accuracy. The proposed approach holds significant potential to improve the timeliness and reliability of wind-related early warning systems, thus contributing to more effective disaster risk reduction strategies.

A Fast Dynamic Assignment Algorithm for Solving Resource Allocation Problems Amalia, Ivanda Zevi; Saikhu, Ahmad; Soelaiman, Rully
JOIN (Jurnal Online Informatika) Vol 6 No 1 (2021)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v6i1.692

The assignment problem is one of the fundamental problems in the field of combinatorial optimization. The Hungarian algorithm can be developed to solve various assignment problems according to each criterion. The assignment problem that is solved in this paper is a dynamic assignment to find the maximum weight on the resource allocation problems. The dynamic characteristic lies in the weight change that can occur after the optimal solution is obtained. The Hungarian algorithm can be used directly, but the initialization process must be done from the beginning every time a change occurs. The solution becomes ineffective because it takes up a lot of time and memory. This paper proposed a fast dynamic assignment algorithm based on the Hungarian algorithm. The proposed algorithm is able to obtain an optimal solution without performing the initialization process from the beginning. Based on the test results, the proposed algorithm has an average time of 0.146 s and an average memory of 4.62 M. While the Hungarian algorithm has an average time of 2.806 s and an average memory of 4.65 M. The fast dynamic assignment algorithm is influenced linearly by the number of change operations and quadratically by the number of vertices.

THE EFFECT OF FACIAL ACCESSORY AUGMENTATION ON THE ACCURACY OF DEEP LEARNING-BASED FACIAL RECOGNITION SYSTEMS Hidayat, Ahmad Nur; Suciati, Nanik; Saikhu, Ahmad
JURTEKSI (jurnal Teknologi dan Sistem Informasi) Vol. 11 No. 3 (2025): Juni 2025
Publisher : Lembaga Penelitian dan Pengabdian Kepada Masyarakat (LPPM) STMIK Royal Kisaran

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33330/jurteksi.v11i3.3846

Abstract: Face recognition based on deep learning has become an important technology in many areas. However, these systems often face challenges in real-world conditions, such as when the face is partially covered by accessories such as masks or glasses. This study aims to evaluate the effect of data augmentation by adding facial accessories (masks, glasses, and a combination of both) and geometric augmentation on the accuracy of face recognition systems. There are three types of datasets used in this method: the original dataset (category 1), the dataset with facial accessories augmentation (category 2), and the dataset with geometric augmentation (category 3). Data augmentation was performed on the training dataset to increase diversity, followed by the face detection process using SCRFD and feature extraction with ArcFace. The model was then trained using Multi-Layer Perceptron (MLP). Based on the results, adding face accessories (category 2) made the model a lot more accurate, hitting 99% accuracy. In category 3, adding geometric features improved accuracy to 91%. Other evaluation metrics, such as precision, recall, and F1-score, also showed improvement after augmentation. This study concludes that facial accessories augmentation is more effective in improving the accuracy and robustness of face recognition models compared to geometric augmentation.Keywords: augmentation; deep learning; face recognition; glasses. Abstrak: Pengenalan wajah berbasis deep learning telah menjadi salah satu teknologi penting dalam berbagai aplikasi. Namun, sistem ini sering kali menghadapi tantangan dalam kondisi dunia nyata, seperti saat wajah tertutup sebagian oleh aksesori seperti masker atau kacamata. Penelitian ini bertujuan untuk mengevaluasi pengaruh augmentasi data dengan menambahkan aksesori wajah (masker, kacamata, dan kombinasi keduanya) serta augmentasi geometris terhadap akurasi sistem pengenalan wajah. Metode yang digunakan melibatkan tiga kategori dataset: dataset asli tanpa augmentasi (kategori 1), dataset dengan augmentasi aksesoris wajah (kategori 2), dan dataset dengan augmentasi geometris (kategori 3). Augmentasi data dilakukan pada dataset pelatihan untuk meningkatkan keberagaman, diikuti dengan proses deteksi wajah menggunakan SCRFD dan ekstraksi fitur dengan ArcFace. Model kemudian dilatih menggunakan Multi-Layer Perceptron (MLP). Hasil penelitian menunjukkan bahwa augmentasi aksesoris wajah (kategori 2) memberikan peningkatan signifikan pada akurasi model, mencapai 99%, sedangkan kategori 3 dengan augmentasi geometris mencapai akurasi 91%. Metrik evaluasi lainnya, seperti precision, recall, dan F1-score, juga menunjukkan peningkatan setelah augmentasi. Penelitian ini menyimpulkan bahwa augmentasi aksesoris wajah lebih efektif dalam meningkatkan akurasi dan ketahanan model pengenalan wajah dibandingkan dengan augmentasi geometris.Kata kunci: augmentasi; deep learning; kacamata; pengenalan wajah.

FACE RECOGNITION USING DEEP NEURAL NETWORKS WITH THE COMBINATION OF DISCRETE WAVELET TRANSFORM, STATIONARY WAVELET TRANSFORM, AND DISCRETE COSINE TRANSFORM METHODS Afrizal Laksita Akbar; Chastine Fatichah; Ahmad Saikhu
JUTI: Jurnal Ilmiah Teknologi Informasi Vol. 18, No. 2, July 2020
Publisher : Department of Informatics, Institut Teknologi Sepuluh Nopember

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j24068535.v18i2.a1000

Personal identification can be done by using face, fingerprint, palm prints, eye’s retina, or voice recognition which commonly called as biometric methods. Face recognition is the most popular and widely used among those biometric methods. However, there are some issues in the implementation of this method: lighting factor, facial expression, and attributes (chin, mustache, or wearing some accessories). In this study, we propose a combination method of Discrete Wavelet Transform and Stationary Wavelet Transform that able to improve the image quality, especially in the small-sized image. Moreover, we also use Histogram Equalization in order to correct noises such as over or under exposure, Discrete Cosine Transform in order to transform the image into frequency domain, and Deep Neural Networks in order to perform the feature extraction and classify the image. A 10-fold cross-validation method was used in this study. As the result, the proposed method showed the highest accuracy up to 92.73% compared to Histogram Equalization up to 80.73%, Discrete Wavelet Transform up to 85.85%, Stationary Wavelet Transform up to 64.27%, Discrete Cosine Transform up to 89.50%, the combination of Histogram Equalization, Discrete Wavelet Transform, and Stationary Wavelet Transform up to 69.77%, and the combination of Stationary Wavelet Transform, Discrete Wavelet Transform, and Histogram Equalization up to 77.39%.

PREDICTION OF MULTIVARIATE TIME SERIES DATA USING ECHO STATE NETWORK AND HARMONY SEARCH Muhammad Muharrom Al Haromainy; Chastine Fatichah; Ahmad Saikhu
JUTI: Jurnal Ilmiah Teknologi Informasi Vol. 19, No. 2, Juli 2021
Publisher : Department of Informatics, Institut Teknologi Sepuluh Nopember

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j24068535.v19i2.a1051

Multivariate time series data prediction is widely applied in various fields such as industry, health, and economics. Several methods can form prediction models, such as Artificial Neural Network (ANN) and Recurrent Neural Network (RNN). However, this method has an error value more significant than the development method of RNN, namely the Echo State Network (ESN). The ESN method has several global parameters, such as the number of reservoirs and the leaking rate. The determination of parameter values dramatically affects the performance of the resulting prediction model. The Harmony Search (HS) optimization method is proposed to provide a solution for determining the parameters of the ESN method. The HS method was chosen because it is easier to implement, and based on other research, the HS method gets the optimum value better than other meta-heuristic methods. The methods compared in this study are RNN, ESN, and ESN-HS. Root Mean Square Error (RMSE) and Mean Absolute Percent Error (MAPE) are used to measure the error rate of forecasting results. ESN got a smaller error value than RNN, and ESN-HS produced a minor error value among the other trials, namely 0.782e-5 for RMSE and 0.28% for MAPE. The HS optimization method has successfully obtained the appropriate global parameters for the ESN prediction model.

Title

Found 31 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 31 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 31 Documents
Search