cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Andi Baso Kaswar
Contact Email
a.baso.kaswar@gmail.com
Phone
+6285656227888
Journal Mail Official
fakhri@diginus.id
Editorial Address
Antang, Makassar, South Sulawesi, Indonesia
Location
Kota makassar,
Sulawesi selatan
INDONESIA
Journal of Deep Learning, Computer Vision and Digital Image Processing
Published by Sakura Digital Nusantara
ISSN : 29868920     EISSN : 29868939     DOI : https://doi.org/10.61255/decoding
Core Subject : Science,
Computer Science & IT
The Journal of Deep Learning, Computer Vision and Digital Image Processing (DECODING), covers all topics of artificial intelligence and soft computing and their applications, including but not limited to: • Neural networks • Reasoning and evolution • Intelligent search • Intelligent planning • Intelligence applications • Computer vision and speech understanding • Multimedia and cognitive informatics • Data mining and machine learning tools, heuristic and AI planning strategies and tools, computational theories of learning • Technology and computing (like particle swarm optimization); intelligent system architectures • Knowledge representation • Bioinformatics • Natural language processing • Automated reasoning • Logic programming • Machine learning • Visual/linguistic perception • Evolutionary and swarm algorithms • Derivative-free optimisation algorithms • Fuzzy sets and logic • Rough sets • Simulated biological evolution algorithms (like genetic algorithm, ant colony optimization, etc) • Multi-agent systems • Data and web mining • Emotional intelligence • Hybridisation of intelligent models/algorithms • Parallel and distributed realisation ofintelligent algorithms/systems • Application in pattern recognition, image understanding, control, robotics and bioinformatics • Application in system design, system identification, prediction, scheduling and game playing • Application in VLSI algorithms and mobile communication/computing systems
Arjuna Subject : Ilmu Komputer (semua kategori) - Kecerdasan Buatan
Articles 63 Documents
 3   4   5   6   7 
Integration of named entity recognition and latent Dirichlet allocation for extracting cyberbullying issues on X Juanda Pratama; Defry Hamdhana; Zara Yunizar
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 2 June 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i2.1528

Abstract

Purpose – The rapid growth of social media Platform X has increased the risk of cyberbullying, which is difficult to detect due to the unstructured nature of textual data. This study proposes an integration of Named Entity Recognition (NER) and Latent Dirichlet Allocation (LDA) to support the extraction of cyberbullying-related social issues.Methods – A total of 2,000 tweets were processed through preprocessing, spaCy-based entity extraction, TF-IDF weighting, and LDA topic modeling. The latent topics generated by LDA were manually mapped into four predefined categories (Bodyshaming, Racism, Gender, and Neutral) and evaluated against researcher-annotated ground truth labels.Findings – Experimental results achieved an overall accuracy of 80%, with F1-scores of 94% for Racism, 93% for Gender, 70% for Bodyshaming, and 63% for Neutral.Research implications – The proposed framework provides practical support for monitoring cyberbullying patterns and assisting policymakers in understanding online social issues.Originality – The originality of this research lies in the sequential integration of NER as an entity-filtering stage prior to LDA, enabling a more comprehensive analysis of cyberbullying discussions than the isolated application of either method.
Comparative Analysis of a Simple CNN and Fine-Tuned ResNet50 for Deepfake Image Detection: Performance and Computational Efficiency Evaluation Rifda Triani Mutmainah; Asti Herliana
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 2 June 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i2.1629

Abstract

Purpose – Deepfake-related cybercrime is an increasingly troubling cybersecurity threat, since such content can now be produced from a single facial photo using freely available face-swapping techniques, while human ability to distinguish real from fake images remains limited (48.2–59%). This study compares a simple CNN and a fine-tuned ResNet50 within an identical, controlled framework.Methods – A controlled experiment compared a simple CNN (trained from scratch) and ResNet50 (two-phase fine-tuning) on 20,000 images from the Kaggle Deepfake and Real Images dataset (80:10:10 split, seed = 42). Candidate images were deduplicated before the data split; zero cross-split duplicates were confirmed. ResNet50 used ResNet-specific preprocessing; the CNN used inputs normalized to [0,1]. Evaluation used accuracy, precision, recall, F1-score, AUC-ROC, and specificity, with a paired McNemar's test as the primary significance measure.Findings – The CNN outperformed ResNet50 on six of seven metrics (accuracy 90.90% vs. 89.80%; AUC-ROC 97.08% vs. 96.52%), while ResNet50 achieved higher recall (93.10% vs. 91.80%). The accuracy difference was not statistically significant (p = 0.200). The CNN was 69.5 times smaller and trained faster (12.2 vs. 16.5 minutes). Neither model showed clear overfitting. Research implications – The findings rest on a single subset, a single split, and a single training run per model, limited to 20 epochs; the potential of ResNet50 with longer training has not been explored.Originality – This study combines accuracy, generalization-related diagnostics, and efficiency in a single controlled comparison, applies a paired test appropriate for a shared test set, and filters duplicates before the data split.
The AI–Gamification Integrated HR Control (AGIHC) Model: A Conceptual Framework for Employee Selection and Placement in Indonesia Sony Putra; Wilda Fasim; Basri; Etty Sri Wahyuni
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 3 September 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i3.1423

Abstract

Purpose – Selection and placement in many organizations remain subjective and inefficient, while gamification’s engagement potential is rarely applied to hiring. This study explains why an integrated approach is needed and proposes the AI–Gamification Integrated HR Control (AGIHC) Model, which couples AI as an analytic engine with gamification as an engagement interface to optimize the HR control system in an emerging-market (Indonesian) context.Methods – A Design Science Research approach was combined with a structured narrative synthesis of Scopus- and Sinta-indexed literature (2020–2026). The model was developed as a designed artifact, with a mixed-methods validation roadmap specified for the next phase.Findings – Based on the synthesized literature rather than on primary measurement in this study, individual primary studies report reductions in time-to-hire and cost-per-hire on the order of one-third; this figure is taken from those cited studies and is not an average computed in the present synthesis. Only 7.3% of gamification studies address recruitment versus 85.4% targeting existing employees. The resulting AGIHC Model specifies three layers: an Input Layer (AI screening plus gamified assessment), a Process Layer (AI matching plus gamified onboarding), and an Output Layer (AI analytics plus a gamified dashboard). Research implications – As a conceptual contribution, the propositions await empirical testing, and the single-country Indonesian framing bounds generalizability. Deployment also depends on data governance, candidate consent, and the explainability of AI scoring.Originality – The study bridges two previously parallel literatures in an integrated AI–gamification HR control framework and provides testable propositions for engagement-driven, fairness-aware selection and placement.
A Conceptual Framework for Digital-Ethical Literacy and Social Responsibility in Islamic Religious Education in the Generative AI Era Ridha Aulia; Siti Suryanih; Aproni; Syahrul Arya Fatahuddin; Jupri
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 2 June 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i2.1551

Abstract

Purpose – This study develops a conceptual framework of digital-ethical literacy and social responsibility for responsible Generative Artificial Intelligence (GenAI) use in Islamic Religious Education by integrating AI literacy, digital pedagogy, Islamic ethics, and digital citizenship.Methods – A conceptual literature review was conducted using a purposive corpus of 110 sources selected for thematic relevance, conceptual contribution, and argumentative strength. The literature was analyzed through thematic-conceptual synthesis to identify recurring constructs and relationships.Findings – The synthesis produced four progressive operational components: AI awareness, critical verification, ethical academic use, and socially responsible digital participation. Islamic moral orientation functions as a cross-cutting normative dimension guiding all components through amanah, tabayyun, adab, ‘adl, responsibility, and maslahah. The framework is progressive but non-rigid and context-sensitive, emphasizing educator guidance, authoritative sources, prior learner knowledge, institutional policy, and topic sensitivity. Research implications – The framework provides a basis for curriculum design, educator guidance, assessment development, institutional policy, and future empirical validation.Originality – The study offers an integrated mechanism linking technological understanding, epistemic verification, academic integrity, Islamic moral judgment, and digital social responsibility in one context-sensitive model.
Perceived AI Utilization and Digital Literacy as Predictors of Government Social Media Communication Effectiveness in Aceh Province Mena Sari
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 2 June 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i2.1612

Abstract

Purpose – This study examines the associations and predictive contributions of perceived AI-supported government communication and digital literacy to perceived public communication effectiveness through official government social media in Aceh Province.Methods – A quantitative cross-sectional survey was conducted among 180 Aceh residents who had accessed or interacted with at least one official government social media account. Data were collected using a five-point Likert-scale questionnaire and analyzed using descriptive statistics, preliminary instrument assessment, regression-assumption testing, and multiple linear regression.Findings – Perceived AI-supported government communication was positively associated with perceived public communication effectiveness (B = 0.389, β = 0.347, p < 0.001). Digital literacy was also positively associated with perceived public communication effectiveness (B = 0.472, β = 0.441, p < 0.001) and showed a relatively larger standardized coefficient, although the difference between coefficients was not formally tested. Together, both predictors explained 52.4% of the variance in perceived public communication effectiveness (R² = 0.524. Research implications – The findings suggest that government institutions in Aceh should combine responsible AI-supported communication practices with staff capability development, human oversight, accessible public information, and initiatives that strengthen citizens’ digital literacy. Because the study used cross-sectional, purposive, and citizen-reported data, these implications should be verified through more representative, longitudinal, and multi-source research.Originality – This study integrates citizens’ perceptions of AI-supported government communication and digital literacy within a single analytical model, providing empirical evidence on their joint relationship with perceived public communication effectiveness in a local-government social media context
Design of an Apperception Strategy to Activate Students’ Prior Knowledge Using Visual Block Programming Aria Sastra Wisesa; Jajang Kusnendar; Muhammad Rafi Valliansyah; Lala Septem Riza
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 3 September 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i3.1633

Abstract

Purpose – This study aims to implement an Activating Prior Knowledge (APK) strategy supported by the OOPify visual block programming tool in Object-Oriented Programming (OOP) learning and to examine students' learning outcomes and learning responses following its implementation.Methods – The study employed the Research and Development (R&D) method, consisting of preliminary study, product development, implementation, and evaluation. Data were collected through observations, interviews, pretest–posttest assessments, and questionnaires. The data were analyzed using the Shapiro–Wilk normality test, the Wilcoxon Signed-Rank Test, Normalized Gain (N-Gain) analysis, and descriptive statistics.Findings – The Wilcoxon Signed-Rank Test showed a statistically significant difference between the pretest and posttest scores (Z = −3.346, p = 0.001). The N-Gain analysis indicated that students with low initial proficiency achieved the highest average N-Gain score of 0.493 (moderate category). In addition, the usability evaluation showed that OOPify obtained an overall usability score of 76.23%, indicating good usability and positive student perceptions. Research implications – These findings provide preliminary evidence that integrating an Activating Prior Knowledge strategy with the OOPify visual block programming platform may support students' conceptual understanding and learning experiences. Because this study employed a one-group pretest–posttest design, the observed differences should not be interpreted as definitive causal effects.Originality – This study integrates the Activating Prior Knowledge strategy through the Know–Want to Know–Learned (KWL) approach and brainstorming activities with the OOPify visual block programming tool to support more meaningful learning of Object-Oriented Programming concepts
Multivariate LSTM versus Simple Baselines for FFB Yield Forecasting at Marihat Plantation: A Feasibility Pilot Rendi Tajemir; Raden Aris Sugianto; Sri Lestari Rahayu
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 2 June 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i2.1715

Abstract

Purpose – This study re-evaluates a multivariate LSTM for forecasting oil palm fresh fruit bunch (FFB) yield at Marihat Plantation after an audit identified a data-export bug in the original dataset. The data were rebuilt and assessed using a leakage-safe chronological design, naive and moving-average baselines, and seed-sensitivity checks.Methods – Monthly FFB production, rainfall, and rainy-day records from plantation reports (2020–2023) were reconstructed and aggregated by planting-year cohort because block-level data were unavailable. Twelve-month windows across 11 cohorts yielded 240 sequences. With an 11-month purge gap, the chronological split comprised 10 training, 20 validation, and 10 test sequences. A two-layer LSTM (64 and 32 units; 20% dropout) was compared with five naive and moving-average baselines and retrained using 10 random seeds.Findings – The LSTM achieved MAPE = 29.73% (MAE = 0.27; RMSE = 0.34 ton/ha) on the test set, outperforming all baselines (MAPE = 74–118%). Across 10 seeds, mean MAPE was 31.79% (SD = 4.59%; worst = 40.60%), remaining below every baseline. However, all test sequences represented December 2023 across 10 cohorts. The observed advantage therefore indicates feasibility, not forecasting superiority across time.Research implications – The LSTM produced consistently lower errors than the baselines across cohorts for one target month, but the 10-sequence, single-month test set does not support general temporal claims. The prototype dashboard may support scenario exploration, although its 12-month forecasts depend on deterministic, unvalidated synthesized inputs.Originality/value – Prompted by a data-quality audit documented in Supplementary S1, this study transparently rebuilds the evidence base and reframes the model as a feasibility pilot rather than a validated solution. Longer, gap-free records covering multiple test months and a validated multi-step procedure are required to establish a dependable advantage.
Integrating Immersive Learning and Interactive Media in Teaching Islamic Cultural History: A Qualitative Case Study Novia Ramadhani; Wahyudin Nur Nasution
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 3 September 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i3.1771

Abstract

Purpose – This exploratory qualitative case study describes the initial implementation of technology-based learning media (Wordwall, video, and PowerPoint) within Islamic Cultural History (ICH) learning at MTsN 2 Deli Serdang.Methods – The study involved five informants: one (ICH) teacher, a principal as a contextual informant, and three eighth-grade students. Data were collected naturally through eight classroom observation sessions, documentation studies, and five in-depth interviews. To comply with ethical standards for protecting minors, all student identities were anonymized, and visual documentation was digitally blurred.Findings – Based on observations and interviews, integrating interactive media reduced student passivity common in conventional lecture methods. Visual quizzes and videos helped direct student focus and facilitated historical recollection. However, given the small qualitative sample size, improvements in students' motivation, critical thinking, and social skills remained situational and did not reach full theoretical saturation. Practical implementation was also constrained by limited lesson time, infocus shortages, and the challenges of large crowd management due to noise during gamified quizzes. Research implications – This study offers context-specific qualitative insights into how interactive tools stimulate classroom engagement in madrasah settings. Originality – The findings reject macro-level generalizations or broad TPACK and self-determination claims, illustrating instead the practical challenges
Development of OOPify as a Visual Block Programming Environment for OOP Learning and Group-Level Practical Performance Naufal Oktavian; Yogi Prasetyo; Muhammad Rafi Valliansyah; Lala Septem Riza
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 3 September 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i3.1634

Abstract

Purpose – Object-Oriented Programming (OOP) is challenging for students because its abstract concepts are difficult to translate into practical programming solutions. This study developed OOPify, a visual block-based learning environment for OOP practical activities, and examined changes in students’ conceptual understanding and group-level practical performance.Methods – The study employed a Research and Development approach using the ADDIE model and a one-group pretest–posttest design. Participants were 34 vocational high school students in West Java, Indonesia. Data were collected through OOP conceptual understanding tests, group-level practical performance observations, expert validation, and student response questionnaires.Findings – OOPify was rated feasible (82.5%) by one expert. The Wilcoxon Signed-Rank Test indicated a significant improvement in OOP conceptual understanding (p < .001), with 70.59% of students achieving moderate-to-high N-Gain and an average N-Gain of 0.50. Group-level practical performance, covering psychomotor, cognitive, and collaborative indicators, averaged 85.22% across Meetings 2 and 3. However, these results represent preliminary observational evidence because formal inter-rater reliability was not assessed.Research implications – Visual block-based learning environments may support OOP learning by reducing syntactic complexity and allowing students to focus on concepts and problem-solving. Generalizability is limited by the single-school context and absence of a control group.Originality – This study integrates visual block programming, OOP practical activities, and group-level practical performance within a single learning environment.
Enhancing Enhancing High School Learning Management Systems Through Explainable AI-Based Behavioral Analytics for Personalized Learning Support Yaya Sudarya Triana; Nurhasanah; Hadi Santoso
Journal of Deep Learning, Computer Vision, and Digital Image Processing Volume 4 Issue 3 September 2026
Publisher : CV. Sakura Digital Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61255/decoding.v4i3.1676

Abstract

Purpose – This study develops an Explainable AI (XAI)-enhanced Learning Analytics framework for analyzing student learning behaviors in a senior high school Learning Management System (LMS) and supporting retrospective performance-risk identification using score-independent behavioral features.Methods – The study analyzed 57,095 LMS quiz activity records from 111 students collected between July 2024 and October 2025. Seven behavioral variables were initially engineered, while Answer Accuracy, Learning Consistency, and Learning Efficiency were excluded from predictive modeling because they are derived from score-related information. The final models used Quiz Frequency, Learning Duration, Activity Level, and Dominant Study-Time Pattern. Random Forest, Decision Tree, and Logistic Regression were evaluated, with SHAP TreeExplainer used for interpretation.Findings – Decision Tree achieved the highest accuracy (73.9%), macro F1-score (0.638), and macro ROC-AUC (0.771) on the 23-student test set. Random Forest achieved 69.6% accuracy, 0.631 macro F1-score, and 0.669 macro ROC-AUC, while Logistic Regression achieved 56.5%, 0.455, and 0.656, respectively. Learning Duration had the highest mean absolute SHAP value.Research Implications – The results indicate that score-independent behavioral features provide a limited-to-moderate signal for retrospective performance classification. Prospective early-warning utility requires temporal validation.Originality – The study integrates XAI with behavioral Learning Analytics while explicitly separating score-independent predictors from the score-derived target to support interpretable educational decision-making

Page 6 of 7 | Total Record : 63

 3   4   5   6   7 

Filter by Year

2023 2026


Reset
Filter By Issues
All Issue Volume 4 Issue 3 September 2026 Volume 4 Issue 2 June 2026 Volume 4 Issue 1 March 2026 Volume 3 Issue 2 September 2025 Volume 3 Issue 1 Maret 2025 Volume 2 Issue 2 September 2024 Volume 2 Issue 1 Maret 2024 Volume 1 Issue 2 September 2023 Volume 1 Issue 1 Maret 2023