IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Enhanced solar panels fault detection approach using lightweight YOLO
Get PDFArtificial intelligence (AI)-driven fault detection improves the reliability of solar energy by reducing the chances of system failures. However, existing single-stage object detection methods excel in accuracy but demand high computational resources, preventing seamless integration into embedded systems. This paper introduces a lightweight approach using YOLOv5, which incorporates a multi-backbone design, specifically tailored for accurate fault detection in solar cells. It evaluates YOLOv5 and TinyYOLOv5. The findings emphasize the effectiveness of YOLOv5l with Ghost backbone, particularly notable for its precision rates of 96% for faulty and 93% for non-faulty instances. Additionally, it showcases commendable mean average precision (mAP) scores, achieving 78% at an intersection over union (IoU) threshold of 0.5 and 72% at an IoU of 0.95. Additionally, YOLOv5_Ghost emerges as the optimal selection, showcasing competitive precision, processing speed of 42.1 giga floating point operations per second (GFLOPS), and remarkable efficiency with 2.4 million parameters. This evaluation underscores the effectiveness of YOLOv5 models, thereby leading to advanced solar energy technology significantly
Automated legal content management system for multi-country integration
Get PDFThis paper presents an automated legal content management system (CMS) designed for multi-country integration, addressing the complex challenges of legal document migration across more than 180 countries while ensuring regulatory compliance and accessibility standards. The system implements a hierarchical four-level architecture, migrating more than 2,740 legal documents with zero data loss incidents through fault-tolerant processing pipelines. The automated portable document format (PDF) migration component demonstrates exceptional efficiency, processing documents 36 times faster compared to manual approaches, while article migration achieves 230 times faster processing speeds. The integrated artificial intelligence (AI)-powered accessibility enhancement system generates contextually appropriate alt text descriptions, allowing organizations to process 10,000 images annually with savings of $14,990. The complete country migration process, covering both PDF and article processing, executes in 30 seconds compared to 56 minutes for manual processing, representing a 112-fold improvement in performance. System scalability demonstrates linear performance characteristics up to more than 5,000 documents with consistent processing metrics while maintaining compliance across diverse regulatory frameworks. These quantitative improvements establish a new paradigm for automated legal content management, providing a scalable foundation for global enterprises managing multi jurisdictional legal documentation requirements
Classification of single origin Indonesian coffee beans using convolutional neural network
Get PDFThis research aims to develop a coffee bean type detection model using convolutional neural networks (CNN), leveraging a dataset of 14,525 images from 116 types of Indonesian coffee beans. Pre-processing steps including resizing, rescaling, and augmentation were applied to improve the dataset quality. The dataset was split into training, validation, and testing sets with proportions of 80%, 10%, and 10%, respectively. Two model development approaches were used: transfer learning with Inception V3 in two scenarios and a model built from scratch. The transfer learning Inception V3 model in scenario 1 achieved the best performance, with a test accuracy of 0.87 and optimal evaluation metrics across precision, recall, and F1-score. This model was fine-tuned using pretrained weights, allowing it to adapt effectively to the coffee bean dataset. The results highlight that transfer learning, especially with Inception V3, provides a robust method for classifying coffee beans, offering potential applications in the coffee industry for improving classification efficiency and accuracy. The study demonstrates how deep learning can enhance the objectivity and precision of coffee bean classification, contributing to greater consistency in product sorting and quality assessment
SaaS reusability assessment using machine learning techniques
Get PDFWith the aid of internet, cloud computing offers hardware and software resources as cloud service. Cloud computing is developing as an effective criterion for reusing. Software-as-a-service (SaaS) is one of the three cloud deployment models that provide on-demand software on a charge basis. A reusability model for SaaS needs to be developed in order to increase its benefits and effectively help end users. A product’s ability to be reused is essential to its easy and effective development. We have presented a software reusability estimation model in this paper. We have assessed the SaaS reusability using machine learning techniques such as adaptive neuro-fuzzy inference system (ANFIS), linear regression, support vector machine (SVM), ensemble, and neural networks. We compared machine learning models using commonality, accessibility, availability, customizability, and efficiency, as the SaaS reusability criteria. The root mean squared error (RMSE), mean squared error (MSE), and mean absolute error (MAE) are used to validate the findings from recommended methodologies with the required level of accuracy. The evaluation's findings have shown that machine learning algorithms yield estimations with a better degree of accuracy, making them more advantageous and practical for SaaS service providers as well as customers
Comparative analysis of machine learning models for fake news detection in social media
Get PDFThe rapid rise of information sharing on social media has amplified the spread of fake news, making its detection increasingly critical. As fake news continues to proliferate, the need for efficient detection mechanisms has become more urgent to protect users from misinformation and disinformation. This paper presents a comparative analysis of multiple machine learning models for detecting text based fake news on social media platforms. Using models such as gradient boosting, XGBoost, and linear support vector classifier (SVC) on the Infor mation Security and Object Technology (ISOT) fake news dataset, the study demonstrates that gradient boosting achieves the highest accuracy of 99.61%, while XGBoost provides a strong balance with 99.59% accuracy and a signifi cantly lower execution time, making it more suitable for real-time applications. These results offer valuable insights into the trade-offs between accuracy and computational efficiency, contributing to the development of more practical de tection systems and future research in the field
Hybrid method for optimizing emotion recognition models on electroencephalogram signals
Get PDFTwo critical factors that need to be studied in emotion recognition are the differences in electroencephalogram (EEG) signal patterns caused by participant characteristics and EEG signals with spatial information. These factors significantly affect the resulting accuracy. The model proposed in this study can consider these factors. This model consists of the modified weighted mean filter method for the basic EEG signal smoothing process, the differential entropy method for the feature extraction process, the relative difference method for the baseline reduction, the 3D cube method for feature representation, and the continuous capsule network method for the classification process. Based on testing on three public datasets, this hybrid method can overcome factors affecting emotion recognition accuracy. This statement is based on the accuracy produced by this model, which outperformed the accuracy validated in previous studies
Detection of partially occluded area in face image using U-Net model
Get PDFOccluded face recognition is important task in computer vision. To complete the occluded face recognition efficiently, first we need to identify the occluded region in face. Identifying the occluded region in face is a challenging task in computer vision. One case of face occlusion is nothing but wearing masks, sunglasses, and scarves. Another case of face occlusion is face is hiding the other objects like books, things, or other faces. In our research, identifying the occluded area which is corona virus disease of 2019 (COVID-19) masked area in face and generate segmentation map. In semantic segmentation, deep learning-based techniques have demonstrated promising outcomes. We have employed one of the deep learning-based U-Net models to generate a binary segmentation map on masked region of a human face. It achieves reliable performance and reducing network complexity. We train our model on MaskedFace-CelebA dataset and accuracy is 97.7%. Results from experiments demonstrate that, in comparison to the most advanced semantic segmentation models, our approach achieves a promising compromise between segmentation accuracy and computing efficiency
Deep lung nodule detection using multi-resolution analysis on computed tomography images
Get PDFThe lung nodule must be detected early because the patient's outcome can be enhanced following the lung cancer diagnosis. The candidate research proposed a novel computer-aided detection system based on multi-resolution technique (MRT) and local Gaussian distribution (LGD) methods to accurately identify and label the lung nodules in a computed tomography (CT) screening image. The research aimed to find all the potential nodule constructs, which combined wavelet and multiscale morphological analysis and then used the LGD method to calculate the Gaussian function parameters for each image block. Subsequently, we calculated the probability that each pixel belongs to a particular institute, which shall be used to achieve lung nodule segmentation reliably. After the segmentation, the research employed a convolutional neural network (CNN) variant to improve the detection performance further. The proposed method attained an accuracy of 0.9958, a sensitivity of 0.7899, a specificity of 0.9994 and an F1-score of 0.8651. The comparison with other methods shows that the proposed method had better detection accuracy than the different methods in terms of lung nodule detection
Hyper-parameters optimized deep feature concatenated network for pediatric pneumonia detection
Get PDFPneumonia, an infection that fills the alveoli of the lung region with pus causes a high rate of chronic illness and fatality amongst children across the globe. The most utilized imaging modality for pediatric pneumonia identification is chest X-rays, whose features are not always readily visible to the naked eye, making it challenging for radiologists to make precise predictions and save lives. Knowing how essential it is to have an early and distinct diagnosis of pneumonia, speeding up or automating the detection process is highly sensible. This article provides a smart, automated system that operates on chest X-ray images and can be successfully utilized for spotting pneumonia. The deep feature concatenation method used by this detection system intends to combine the outcomes of three effective pre-trained models to confirm the reliability of the suggested approach. To obtain its optimal performance, the hyper-parameters are demonstrated using a trial-and-error approach that surpasses existing models with 99.68% accuracy for the early diagnosis of pneumonia. A real-time data sample test is conducted on the proposed pneumonia detection model to evaluate its robustness
Comparison of deep learning models: CNN and VGG-16 in identifying pornographic content
Get PDFIn 2020, a total of 59,741 websites were blocked by the Indonesian government due to containing negative content, including pornography, with 14,266 websites falling into this category. However, these blocked websites could still be accessed by the public using virtual private networks (VPNs). This prompted the research idea to quickly identify pornographic content. This study aims to develop a system capable of identifying websites suspected of containing pornographic image content, using a deep learning approach with convolutional neural network (CNN) and visual geometry group 16 (VGG-16) model. The two models were then explored comprehensively and holistically to determine which model was most effective in detecting pornographic content quickly. Based on the findings of the comparison between testing the CNN and VGG-16 models, research results showed that the best test results were obtained in the eighth experiment using the CNN model at an epoch value level of 50 and a learning rate of 0.001 of 0.9487 or 94.87%. This can be interpreted that the CNN model is more effective in detecting pornographic content quickly and accurately compared to using the VGG-16 model