Claim Missing Document
Check
Articles

Found 4 Documents
Search

Exploring DQN-Based Reinforcement Learning in Autonomous Highway Navigation Performance Under High-Traffic Conditions Sandy Nugroho; De Rosal Ignatius Moses Setiadi; Hussain Md Mehedul Islam
Journal of Computing Theories and Applications Vol. 1 No. 3 (2024): JCTA 1(3) 2024
Publisher : Universitas Dian Nuswantoro

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62411/jcta.9929

Abstract

Driving in a straight line is one of the fundamental tasks for autonomous vehicles, but it can become complex and challenging, especially when dealing with high-speed highways and dense traffic conditions. This research aims to explore the Deep-Q Networking (DQN) model, which is one of the reinforcement learning (RL) methods, in a highway environment. DQN was chosen due to its proficiency in handling complex data through integrated neural network approximations, making it capable of addressing high-complexity environments. DQN simulations were conducted across four scenarios, allowing the agent to operate at speeds ranging from 60 to nearly 100 km/h. The simulations featured a variable number of vehicles/obstacles, ranging from 20 to 80, and each simulation had a duration of 40 seconds within the Highway-Env simulator. Based on the test results, the DQN method exhibited excellent performance, achieving the highest reward value in the first scenario, 35.6117 out of a maximum of 40, and a success rate of 90.075%.
Enhanced Vision Transformer and Transfer Learning Approach to Improve Rice Disease Recognition Rahadian Kristiyanto Rachman; De Rosal Ignatius Moses Setiadi; Ajib Susanto; Kristiawan Nugroho; Hussain Md Mehedul Islam
Journal of Computing Theories and Applications Vol. 1 No. 4 (2024): JCTA 1(4) 2024
Publisher : Universitas Dian Nuswantoro

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62411/jcta.10459

Abstract

In the evolving landscape of agricultural technology, recognizing rice diseases through computational models is a critical challenge, predominantly addressed through Convolutional Neural Networks (CNN). However, the localized feature extraction of CNNs often falls short in complex scenarios, necessitating a shift towards models capable of global contextual understanding. Enter the Vision Transformer (ViT), a paradigm-shifting deep learning model that leverages a self-attention mechanism to transcend the limitations of CNNs by capturing image features in a comprehensive global context. This research embarks on an ambitious journey to refine and adapt the ViT Base(B) transfer learning model for the nuanced task of rice disease recognition. Through meticulous reconfiguration, layer augmentation, and hyperparameter tuning, the study tests the model's prowess across both balanced and imbalanced datasets, revealing its remarkable ability to outperform traditional CNN models, including VGG, MobileNet, and EfficientNet. The proposed ViT model not only achieved superior recall (0.9792), precision (0.9815), specificity (0.9938), f1-score (0.9791), and accuracy (0.9792) on challenging datasets but also established a new benchmark in rice disease recognition, underscoring its potential as a transformative tool in the agricultural domain. This work not only showcases the ViT model's superior performance and stability across diverse tasks and datasets but also illuminates its potential to revolutionize rice disease recognition, setting the stage for future explorations in agricultural AI applications.
Hiragana Character Classification Using Convolutional Neural Networks Methods based on Adam, SGD, and RMSProps Optimizer Ibnu Utomo Wahyu Mulyono; Yupie Kusumawati; Ajib Susanto; Christy Atika Sari; Hussain Md Mehedul Islam; Mohamed Doheir
Scientific Journal of Informatics Vol. 11 No. 2: Regular Issue
Publisher : Universitas Negeri Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15294/sji.v11i2.2313

Abstract

Purpose: Hiragana image classification poses a significant challenge within the realms of image processing and machine learning. Despite advances, achieving high accuracy in Hiragana character recognition remains elusive. In response, this research attempts to enhance recognition precision through the utilization of a Convolutional Neural Network (CNN). Specifically, the study explores the efficacy of three distinct optimizers like Adam, Stochastic Gradient Descent with Momentum (SGDM), and RMSProp in improving Hiragana character recognition accuracy. Methods: This research adopts a systematic approach to evaluate the performance of a Convolutional Neural Network (CNN) in the context of Hiragana character recognition. A meticulously prepared dataset is utilized for in-depth testing, ensuring robustness and reliability in the analysis. The study focuses on assessing the effectiveness of three prominent optimization methods: Stochastic Gradient Descent (SGD), RMSProp, and Adam. Result: The results of the model performance evaluation show that the highest accuracy was obtained from the RMSP optimizer with an F1-Score reaching 99.70%, while the highest overall accuracy was 99.87% with the Adam optimizer. The analysis is carried out by considering important metrics such as precision, recall, and F1-Score for each optimizer. Novelty: The performance results of the developed model are compared with previous studies, confirming the effectiveness of the proposed approach. Overall, this research makes an important contribution to Hiragana character recognition, by emphasizing the importance of choosing the right optimizer in improving the performance of image classification models.
Hybrid Quantum Representation and Hilbert Scrambling for Robust Image Watermarking Christy Atika Sari; Abdussalam Abdussalam; Eko Hari Rachmawanto; Hussain Md Mehedul Islam
Scientific Journal of Informatics Vol. 11 No. 4: Special Issue
Publisher : Universitas Negeri Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15294/sji.v11i4.10140

Abstract

Purpose: This work aims to apply Quantum Hilbert Scrambling to enhance the security and integrity of image watermarking without affecting visual quality degradation. Further conception of the surveyed methods could result in a very good solution to conventional methods of watermarking in solving some problems of digital image security and integrity with new concepts of quantum computing. Methods: The paper reviews Quantum Hilbert Scrambling, whose computational complexity is . The process involves encoding the image into a quantum state, permuting qubits by the Hilbert curve, and embedding a watermark using quantum gates. Result: The quantitative performance evaluation metrics, like Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index (SSIM), have shown high Peak Signal to Noise Ratio (PSNR) values from 56.13 dB to 57.87 dB and Structural Similarity Index (SSIM) from 0.9985 to 0.9990, correspondingly. This justifies the fact that the quality degradation is very slight and the fine details of the structure are well maintained. Novelty: The proposed method uniquely integrates quantum computing with traditional watermarking steps for a secure and effective approach in digital watermarking. Further development should focus on improving the quantum circuit regarding computation efficiency, extending the applicability of the method to a wide range of images, and various situations in watermarking, and finding hybrid approaches by combining quantum and classical approaches towards better performance and scalability.