cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
-
Contact Email
-
Phone
-
Journal Mail Official
-
Editorial Address
-
Location
Kota denpasar,
Bali
INDONESIA
Jurnal Nasional Pendidikan Teknik Informatika (JANAPATI)
Published by Universitas Pendidikan Ganesha
ISSN : 20898673     EISSN : 25484265     DOI : -
Core Subject : Science, Education, Engineering,
Computer Science & IT Education Engineering
Jurnal Nasional Pendidikan Teknik Informatika (JANAPATI) is a collection of scientific articles in the field of Informatics / ICT Education widely and the field of Information Technology, published and managed by Jurusan Pendidikan Teknik Informatika, Fakultas Teknik dan Kejuruan, Universitas Pendidikan Ganesha. JANAPATI first published in 2012 and will be published three times a year in March, July, and December. This journal is expected to bridge the gap between understanding the latest research Informatika. In addition, this journal can be a place to communicate and enhance cooperation among researchers and practitioners.
Arjuna Subject : -
Articles 696 Documents
 66   67   68   69   70 
Integrated Dual Layer Machine Unlearning Using Reverse Gradient Descent and Image Obfuscation Nova Eka Budiyanta; Lukas Lukas; Eugenius Kau Suni
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.105824

Abstract

This study proposes a Hybrid Machine Unlearning framework that combines Reverse Gradient Descent (RGD) and Image Obfuscation to enable selective and privacy-preserving forgetting in deep neural networks without full retraining. The proposed method introduces a dual-layer mechanism that operates at both the model level, by reversing gradient updates to remove target class contribution, and the data level, by applying visual obfuscation to eliminate identifiable features prior to unlearning. Experiments were conducted on a subset of the Labeled Faces in the Wild (LFW) dataset containing 96 identity classes, evaluated under two unlearning scenarios with 5 and 10 target classes. Quantitative results show that pure RGD achieved values between 0.10–0.19 and 0.03–0.06 for the respective scenarios, while the hybrid RGD+Obfuscation configuration strengthened the forgetting effect up to 0.82 on MobileNetV2 and 0.26 on ResNet architectures. Model stability remained high, with below 0.27, indicating minimal degradation of non-target representations. The hybrid method achieved an unlearning time of 2–6s per epoch, with an additional 50–65% computational cost due to blurring and pixelation operations yet maintained an 80–90% reduction in total runtime compared to full retraining. These results demonstrate that the proposed hybrid approach effectively enhances forgetting strength while maintaining retention stability and computational efficiency. The method provides a scalable and resource-efficient solution for privacy-aware continual learning and AI lifecycle management, where fast and controlled machine unlearning is required without compromising model integrity.
A Value-Based Structural Model of Generative AI Adoption for Vocational Teachers Krismiyati Krismiyati; Rudy Latuperissa; Hanita Yulia
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.107130

Abstract

This study develops and empirically tests a value-based structural model of Generative AI adoption among vocational teachers, situating technology acceptance within a techno-pedagogy perspective. Extending the Unified Theory of Acceptance and Use of Technology (UTAUT), the model reconceptualizes performance expectancy, effort expectancy, perceived enjoyment, and computer self-efficacy as cognitive–affective antecedents of perceived AI value. Rather than treating adoption as a linear, intention-driven process, this study positions perceived AI value as the central evaluative mechanism shaping attitudes, behavioral intentions, and actual AI use in instructional contexts. A quantitative explanatory design was employed, involving 110 vocational teachers with prior experience using Generative AI tools. Data were collected using a 5-point Likert-scale instrument and analyzed through Partial Least Squares Structural Equation Modeling (PLS-SEM). The findings indicate that perceived AI value is the strongest predictor of both attitude and actual AI usage, while behavioral intention does not significantly predict usage, revealing an intention–behavior gap. Facilitating conditions were also found to be non-significant predictors of actual use. These results suggest that Generative AI adoption in vocational education is primarily value-driven rather than structurally enforced. The study contributes to the techno-pedagogy literature by demonstrating that utilitarian and hedonic value perceptions can directly activate the use of pedagogical technology, thereby extending UTAUT with a value-centered evaluative pathway. Practical implications emphasize the need for pedagogically grounded AI training that enhances perceived instructional relevance, usability, and experiential value, thereby fostering sustainable AI integration in vocational teaching environments.
ML-CAA-BLIP: Multi-Label Cultural-Aware Adapter for Balinese Carving Image Captioning I Putu Bagus Gede Prasetyo Raharja; Anny Yuniarti
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.108404

Abstract

Pre-trained vision-language models such as BLIP have achieved remarkable success in general image captioning tasks. However, their performance on domain-specific applications, particularly cultural heritage documentation, remains limited due to the lack of specialized knowledge and the inability to handle multi-label cultural categories. Full fine-tuning of these large models is computationally expensive and risks catastrophic forgetting, while standard adapter-based methods treat all images uniformly without considering domain-specific class characteristics. This study proposes ML-CAA-BLIP (Multi-Label Cultural-Aware Adapter for BLIP), a novel parameter-efficient adaptation method for Balinese carving image captioning. The proposed method introduces class-specific scaling parameters for each cultural motif category (Barong, Punggel, Keketusan, Gajah, Goak, Cina, and Daun) and employs a learned importance-weighted fusion mechanism to handle multi-label inputs where images contain multiple artistic styles. Experiments conducted on the BaliCarving dataset comprising 2,181 images demonstrate that ML-CAA-BLIP achieves the best BLEU-4 score of 0.2718 (+52.4% improvement over Base BLIP) and ROUGE-L score of 0.5835 (+15.2% improvement) while adding only 903 trainable parameters. The model also shows competitive performance on other metrics including METEOR and BERTScore. These results indicate that cultural-aware adaptation significantly improves domain-specific image captioning while maintaining parameter efficiency, contributing to the digital preservation of Balinese cultural heritage
Fine-Tuning Text-to-Image Diffusion Models with LoRA for Generative Synthesis of Balinese Endek Textile Patterns I Ketut Adi Purnawan; I Komang Gede Jefri Suparjana
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.111817

Abstract

This study investigates the adaptation of a text-to-image diffusion model to generate patterns inspired by traditional Balinese Endek textiles. The objective is to evaluate whether Low-Rank Adaptation (LoRA) can improve the model’s ability to capture structural characteristics of Endek motifs, including repetitive geometric forms, ornamental symmetry, and stylized flora and fauna elements. A curated dataset of 687 Endek textile images was compiled and categorized into four motif classes: geometric, flora, fauna, and decorative. Each image was paired with a concise Indonesian textual description to guide prompt-conditioned image generation. The diffusion model was fine-tuned using LoRA applied to selected attention layers, enabling efficient domain adaptation with a limited number of trainable parameters. Training was conducted for 4,300 steps, showing stable convergence with a minimum loss of 0.00218, maximum loss of 0.75465, and mean loss of 0.12674 ± 0.10169. Quantitative evaluation using Fréchet Inception Distance (FID), Learned Perceptual Image Patch Similarity (LPIPS), Inception Score (IS), and CLIP-based similarity demonstrates consistent improvements after LoRA fine-tuning. The average FID decreased by 29.68%, improving from 394.21 (baseline) to 276.05 (LoRA). Category-level improvements include geometric (423.32 to 278.99), flora (432.37 to 296.23), fauna (379.00 to 273.13), and decorative motifs (342.14 to 255.84). LPIPS scores also decreased across categories, indicating higher perceptual similarity to real textile patterns. CLIP similarity scores ranged between 26–30, confirming strong alignment between generated images and textual prompts. These results demonstrate that LoRA provides an effective and parameter-efficient approach for adapting diffusion models to domain-specific textile pattern generation, particularly for culturally specific datasets such as Balinese Endek textiles.
Hybrid Residual UNet with Triplet Embedded Metric Learning for Low Light Tuberculosis Bacilli Segmentation Sari Ayu Wulandari; I Ketut Eddy Purnama; Eko Mulyanto Yuniarno; Mauridhi Hery Purnomo
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.112209

Abstract

Automated segmentation of Mycobacterium tuberculosis bacilli in Ziehl Neelsen stained sputum smears is essential for scalable tuberculosis (TB) screening, particularly in low resource settings where heterogeneous staining and poor illumination degrade image quality. Most encoder and decoder models such as Unet rely on overlap based supervision and lack embedding level discrimination, leading to feature confusion between bacilli and staining artifacts under low light conditions. To address this limitation, we propose a Hybrid Residual Unet with Triplet Embedded Metric Learning (RTL), which incorporates margin based metric supervision at the residual bottleneck using structured anchor, positive, and negative sampling and a joint Dice, binary cross entropy, and triplet objective. Evaluated on the DDS1 dataset with illumination stratified analysis, RTL outperformed Unet, Resunet, triplet based baselines, and Transunet, achieving higher Dice and mIoU, lower margin violation rates, and significantly improved embedding separability (p < 0.05). RTL also showed reduced performance variance across illumination subsets, indicating improved robustness to domain shift and more reliable bacilli delineation, which can support downstream components of automated TB microscopy workflows (detection, counting and slide level grading).
A Systematic Literature Review of Retrieval-Augmented Generation Implementation for Enhancing Large Language Models in Education I Ketut Resika Arthana; Nyoman Gunantara; Made Sudarma; Made Sukarsa
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 15 No. 1 (2026)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v15i1.112281

Abstract

The rapid advancement of Large Language Models (LLM) has led to the creation of increasingly adaptive intelligent learning systems. However, many educational implementations of LLMs still rely on internal model knowledge without sufficient grounding in reliable external sources, which may reduce response accuracy, contextual relevance, and trustworthiness. One approach proposed to address this limitation is Retrieval-Augmented Generation (RAG), which combines the LLM’s generative capabilities with external information retrieval systems. Nevertheless, evidence regarding how RAG has been implemented, optimized, and evaluated in educational contexts remains fragmented. This study aimed to evaluate the RAG implementation in supporting LLM performance in learning environments by examining (1) the most frequent types of learning activities involving RAG, (2) the RAG implementation effectiveness in improving response quality, (3) the optimization techniques used to improve RAG results, and (4) the challenges and opportunities faced in its integration in education. The study was conducted systematically using the PICO framework, drawing on articles from the Scopus and IEEE databases, spanning the period from 2021 to 2025. The analysis of 50 studies revealed that RAG is most applied in the contexts of question answering, personalized learning, and tutoring, with significant improvements in the aspects of accuracy, relevance, and personalization of responses. However, not all studies explicitly reported the implementation of optimization techniques to RAG. These techniques comprised knowledge injection, prompt engineering, and query expansion. Challenges remain in reasoning, retrieval accuracy, and system integration, but there is strong potential for developing more adaptive and contextualized learning systems. This review recommends stronger retrieval optimization, better alignment with pedagogical objectives, and broader evaluation using educational outcome measures to maximize the impact of RAG-enhanced LLMs in digital education.

Page 70 of 70 | Total Record : 696

 66   67   68   69   70 

Filter by Year

2012 2026


Reset
Filter By Issues
All Issue Vol. 15 No. 2 (2026) Vol. 15 No. 1 (2026) Vol. 14 No. 3 (2025) Vol. 14 No. 2 (2025) Vol. 14 No. 1 (2025) Vol. 13 No. 3 (2024) Vol. 13 No. 2 (2024) Vol. 13 No. 1 (2024) Vol. 12 No. 3 (2023) Vol. 12 No. 2 (2023) Vol. 12 No. 1 (2023) Vol. 11 No. 3 (2022) Vol. 11 No. 2 (2022) Vol. 11 No. 1 (2022) Vol 11, No 1 (2022) Vol. 10 No. 3 (2021) Vol. 10 No. 2 (2021) Vol 10, No 2 (2021) Vol 10, No 1 (2021) Vol. 10 No. 1 (2021) Vol 9, No 3 (2020) Vol. 9 No. 3 (2020) Vol 9, No 2 (2020) Vol. 9 No. 2 (2020) Vol. 9 No. 1 (2020) Vol 9, No 1 (2020) Vol. 8 No. 3 (2019) Vol 8, No 3 (2019) Vol 8, No 2 (2019) Vol. 8 No. 2 (2019) Vol. 8 No. 1 (2019) Vol 8, No 1 (2019) Vol 8, No 1 (2019) Vol 7, No 3 (2018) Vol. 7 No. 3 (2018) Vol 7, No 3 (2018) Vol. 7 No. 2 (2018) Vol 7, No 2 (2018) Vol. 7 No. 1 (2018) Vol 7, No 1 (2018) Vol 7, No 1 (2018) Vol 6, No 3 (2017) Vol 6, No 3 (2017) Vol. 6 No. 3 (2017) Vol 6, No 2 (2017) Vol. 6 No. 2 (2017) Vol 6, No 2 (2017) Vol. 6 No. 1 (2017) Vol 6, No 1 (2017) Vol 6, No 1 (2017) Vol 5, No 3 (2016) Vol. 5 No. 3 (2016) Vol 5, No 3 (2016) Vol. 5 No. 2 (2016) Vol 5, No 2 (2016) Vol 5, No 2 (2016) Vol 5, No 1 (2016) Vol 5, No 1 (2016) Vol. 5 No. 1 (2016) Vol 4, No 3 (2015) Vol. 4 No. 3 (2015) Vol 4, No 3 (2015) Vol 4, No 2 (2015) Vol 4, No 2 (2015) Vol. 4 No. 2 (2015) Vol 4, No 1 (2015) Vol. 4 No. 1 (2015) Vol 4, No 1 (2015) Vol 3, No 3 (2014) Vol. 3 No. 3 (2014) Vol 3, No 3 (2014) Vol. 3 No. 2 (2014) Vol 3, No 2 (2014) Vol 3, No 2 (2014) Vol 3, No 1 (2014) Vol. 3 No. 1 (2014) Vol 3, No 1 (2014) Vol 2, No 3 (2013) Vol. 2 No. 3 (2013) Vol 2, No 3 (2013) Vol. 2 No. 2 (2013) Vol 2, No 2 (2013) Vol 2, No 2 (2013) Vol 2, No 1 (2013) Vol 2, No 1 (2013) Vol. 2 No. 1 (2013) Vol 1, No 3 (2012) Vol 1, No 3 (2012) Vol. 1 No. 3 (2012) Vol. 1 No. 2 (2012) Vol 1, No 2 (2012) Vol 1, No 2 (2012) Vol 1, No 1 (2012) Vol 1, No 1 (2012) Vol. 1 No. 1 (2012) More Issue