CACandra Ahmadi

Candra Ahmadi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Candra Ahmadi has a total of 13 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Detecting AI-Generated Images Using Facial Similarity and Feature Extraction for Digital Security

    by Man-Ling Liao, Jiann-Liang Chen, Candra Ahmadi

    Artificial intelligence (AI) has revolutionized digital imagery, enabling the creation of highly realistic AI-generated images. This advancement, however, poses risks, such as fraud and misinformation. A critical gap exists in effectively distinguishing between AI-generated and authentic human images. Here, we propose a robust detection system utilizing facial feature extraction and similarity analysis. Our approach involves a three-stage process, including facial feature extraction, similarity assessment, and final verification using the EfficientNet model. The system achieved an accuracy rate of 96.4%, demonstrating its effectiveness in differentiating between AI-generated and real images. These findings offer significant implications for improving digital security and mitigating the risks posed by AI-generated content.


  • Open Access

    Enhancing Cybersecurity With P-Code Analysis and XGBoost: A Novel Approach for Malicious VBA Macro Detection in Office Documents

    by Candra Ahmadi, Jiann-Liang Chen, Yi-Cheng Lai

    In the evolving landscape of cybersecurity, the prevalence of malicious Visual Basic for Applications (VBA) macros embedded in Office documents presents a formidable challenge. These macros, while integral to automation, have become potent vehicles for cyber-attacks, necessitating advanced detection techniques. This study introduces a comprehensive framework employing P-Code Analysis and XGBoost, a leading-edge machine learning algorithm, to address this issue. The proposed solution synergizes static analysis of VBA source code with dynamic P-Code structural analysis, enhanced by Natural Language Processing (NLP) techniques for effective feature extraction. By integrating these methodologies, our model adeptly distinguishes between benign and malicious macros, achieving an unprecedented detection accuracy of 98.70% and an F1-score of 98.81% in rigorous testing environments. The core contribution of this research lies in its innovative approach to malicious macro detection, offering a robust framework that significantly improves upon existing methods. Additionally, the utilization of XGBoost for machine learning analysis introduces a novel application in cybersecurity defenses against macro-based threats. The results underscore the efficacy of combining P-Code analysis with machine learning for cybersecurity, marking a significant stride in the detection of sophisticated cyber threats. This study not only advances the domain of cybersecurity but also lays the groundwork for future research, advocating for the exploration of further optimizations and the adaptation of our model to combat evolving attack vectors. Recommended terms: Cybersecurity, Malicious VBA Macro Detection, P-Code Analysis, XGBoost, Machine Learning.


  • Zero-DCE xT: A Computational Approach to Addressing Low-Light Image Enhancement Challenges

    by Candra Ahmadi, Jiann-Liang Chen, Yu-Yi Hsiao

    In the dynamic field of digital image processing, effectively enhancing low-light images presents a significant challenge. This study narrows its focus to the preservation of natural aesthetics in low-light enhancement. The knowledge gap identified is the need for an approach that balances visibility improvement with authentic color retention without computational burden. We introduce the Zero-DCE xT model, an evolution of the Zero-DCE framework, optimized for structure and function in low-light conditions. Our experimental work shows the Zero-DCE xT model’s capability to significantly reduce prediction errors and improve image quality, achieving high scores in PSNR, VIF, and SSIM across various datasets. Notably, Zero-DCE xT combines visual appeal with computational efficiency, making it ideal for real-time applications. The implications of our findings suggest that the Zero-DCE xT model sets a new standard for low-light image processing, marrying robust performance with economical computation, and paving the way for future research in advanced imaging systems.


  • Hybrid Data Collection System for IoT Based Solar Powered Drip Irrigation system using LoRa WAN

    by Wirarama Wedashwara, I Komang Damar Jaya, Made Sutha Yadnya, Candra Ahmadi, Tatang Mulyana

    Outdoor data communication using the internet in Internet of Things (IoT) applications has a risk of interference caused by environmental conditions. Providing an access point for outdoor applications requires more electricity. The proposed research aimed to optimize data communication between local storage, internet and wireless sensor network (WSN). A hybrid data collection system for an IoT-based solar drip irrigation system using LoRa WAN. The Hybrid system in this study consists of a WAN for local control and data collection using LoRa and uploading to the internet. Apart from storing data in indoor and internet local storage, the system allows local users such as farmers to access a web-based user interface without internet. The WSN system in research uses nRF from previous studies. The blackbox testing shows valid results for nRF, LoRa and HTTP GET data communication series to local and online web server interfaces. The results of a data communication loop programming produces a total average time of 2469ms and a delay of 3000ms on outdoor devices. The indoor device uses 24272ms and a delay of 5000ms for one loop. The Quality of Service (QoS) evaluation showed that the lowest PDR archived by nRF with 97.285%. In other side the longest delay is archived by LoRa with 936ms. Through solar power testing, the overall QoS decreases when the light intensity is below 101 lux, and the solar panel voltage is below 0.121v. But the decrease doesn't seem significant, especially in LoRa with the lowest value of 98.378% and 934ms delay. Meanwhile, the lowest delay was achieved by nRF with 247.6ms. The lowest PDR was archieved by HTTP GET with 96.7638%. Meanwhile, the longest delay was archieved by LoRa 876.6ms.


  • Open Access

    IoT-Based Smart Village Transaction System Using RFID and Load Cell Modules

    by Candra Ahmadi, Wirarama Wedashwara, Ni Nyoman Harini Puspita, Jiann-Liang Chen

  • Internet of Things based Smart Energy Audit using Evolutionary Fuzzy Association Rule Mining

    by Wirarama Wedashwara, Candra Ahmadi, I Wayan Agus Arimbawa, I Gede Eka Wiantara Putra

    Energy audits are investigations, studies and examination of power vitality utilization, with the point of proficiency without lessening the exhibition of the system. The paper presented Internet of Things (IoT) based Smart Energy Audit by implementing electrical data collection using Wireless Sensor Network (WSN) and Decision Support System (DSS) using Evolutionary Fuzzy Association Rule Mining (EFARM). The developed system aims to collects data using IoT nodes and summarise data pattern using EFARM in interpretation of Fuzzy Rules and Tree Based Evolutionary Computation (EC). The evaluation performed in five rooms within a week and shown EFARM capable to: Interpreted patterns of electricity consumption in load samples by achieving a high average, confident and score; Compare consumption patterns between measurement areas and the type of measured load; Summarise the rules similarity between areas as general rules and dissimilarities that only exist in some areas as a specific rule, using singletree interpretation; Interpretation of comparison between power consumption patterns with measurable performance, namely temperature and light intensity.