IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
A comparative study of natural language inference in Swahili using monolingual and multilingual models
Get PDFRecent advancements in large language models (LLMs) have led to opportunities for improving applications across various domains. However, existing LLMs fine-tuned for Swahili or other African languages often rely on pre-trained multilingual models, resulting in a relatively small portion of training data dedicated to Swahili. In this study, we compare the performance of monolingual and multilingual models in Swahili natural language inference tasks using the cross-lingual natural language inference (XNLI) dataset. Our research demonstrates the superior effectiveness of dedicated Swahili monolingual models, achieving an accuracy rate of 69%. These monolingual models exhibit significantly enhanced precision, recall, and F1 scores, particularly in predicting contradiction and neutrality. Overall, the findings in this article emphasize the critical importance of using monolingual models in low-resource language processing contexts, providing valuable insights for developing more efficient and tailored natural language processing systems that benefit languages facing similar resource constraints
Performance analysis of a neuromodel for breast histopathology decision support system
Get PDFBreast cancer detection and diagnosis are crucial in reducing mortality rates among women globally. This research article explores an artificial intelligence technique for early breast cancer detection, aiding doctors in making informed decisions for improved patient management. The study employs histopathological analysis of breast tissue microscopically to detect abnormalities, with the aim of categorizing normal tissue, benign lesions, in situ carcinoma, and invasive carcinoma. The proposed technique utilizes an artificial neural network trained using the resilient backpropagation algorithm (RP_ANN). The study further compares the observed performance with those of three other algorithms, including gradient descent algorithm (GDA_ANN), Levenberg-Marquardt algorithm (LM_ANN), and layer sensitivity-based (LSB_ANN) algorithm based on various evaluation metrics. RP_ANN and LSB_ANN demonstrated superior performance, with high validation and training variance accounted for (VAF) and low root mean squared error (RMSE). The results underscore the potential of deep learning-based algorithms for improving breast cancer detection, promising better patient outcomes and enhanced diagnostic accuracy
An efficient load-balancing in machine learning-based DC-DC conversion using renewable energy resources
Get PDFThis paper introduces the machine learning-based DC-DC conversion algorithm (ML-DC2A), a pioneering machine learning (ML) approach designed to enhance load-balancing in DC-DC conversion systems powered by renewable energy sources. Traditional control strategies, such as pulse-width modulation (PWM), maximum power point tracking (MPPT), and basic voltage and current controls, are foundational yet often fall short in adapting to the rapid fluctuation’s characteristic of renewable energy supply. The ML-DC2A optimizes crucial performance indicators including conversion efficiency, reliability, adaptability to energy supply variability, and response time to changing loads. By leveraging predictive analytics and adaptive algorithms, it dynamically manages the conversion process, offering superior performance over traditional techniques. A notable drawback of conventional methods is their inability to anticipate and adjust to real-time changes in energy availability and demand, leading to inefficiencies and potential system instability. The proposed ML-DC2A addresses these challenges by incorporating a sophisticated ML framework that predicts future energy scenarios and adaptively adjusts system parameters to maintain optimal performance. Initial results highlight the transformative potential of integrating ML into renewable energy conversion systems, promising significantly enhanced efficiency and system resilience, thus marking a significant step forward in sustainable energy management
A reinforcement learning paradigm for Vietnamese aspect-based sentiment analysis
Get PDFThis paper presents the task of aspect-based sentiment analysis (ABSA) that recognizes the sentiment polarity associated with each aspect of entities discussed in customers’ reviews, focusing on a low-resourced language, Vietnamese. Unlike conventional classification approaches, we leverage reinforcement learning (RL) techniques by formulating the task as a Markov decision process. This approach allows an RL agent to handle the hierarchical nature of ABSA, sequentially predicting entities, aspects, and sentiments by exploiting review features and previously predicted labels. The agent seeks to discover optimal policies by maximizing cumulative long-term rewards through accurate entity, aspect, and sentiment predictions. The experimental results on public Vietnamese datasets showed that the proposed approach yielded new state of the art (SOTA) results in both hotel and restaurant domains. Using the best model, we achieved an improvement of 1% to 3% in the F1 scores for detecting aspects and the corresponding sentiment polarity
Lightweight mutual authentication protocol for resource-constrained radio frequency identification tags with PRINCE cipher
Get PDFRadio frequency identification (RFID) is a key technology for the internet of things (IoT), with widespread applications in the commercial, healthcare, enterprise, and community sectors. However, privacy and security concerns remain with RFID systems. This manuscript presents a novel RFID-based mutual authentication protocol (MAP) using the PRINCE cipher to address these concerns. The proposed MAP leverages a PRINCE cipher architecture capable of both encryption and decryption based on a mode signal. It performs five encryption and two decryption processes during tag and reader mutual authentication, with updated seed values ensuring synchronization and secure data communication. The PRINCE cipher implementation utilizes less than 1% of slices, operates at 226 MHz with a latency of 3.5 clock cycles (CC), and has a throughput of 4.125 Gbps. The complete RFID-based MAP consumes 721 mW of power, occupies 2% of the chip area, and achieves a latency of 35.5 CC and a throughput of 262 Mbps. This represents a 25% reduction in latency, a 40% increase in throughput, and a 30% decrease in execution time compared to existing MAP approaches. The findings demonstrate the potential of the proposed MAP to enhance latency, throughput, and execution time, offering a promising solution for secure and efficient RFID authentication
Optimizing diabetes prediction: unveiling patient subgroups through clustering
Get PDFDiabetes is a significant global health concern, leading to numerous deaths annually and affecting many individuals who remain undiagnosed. As its prevalence rises, the importance of early detection becomes increasingly vital. The rising diabetes epidemic demands data-driven strategies to catch health problems sooner and identify them clearly. This study utilizes the Pima Indians diabetes dataset (PIDD) to compare three powerful clustering schemes such as k-means, fuzzy C-means, and hierarchical. Uncontrolled diabetes, arising from the body's struggle to manage blood sugar due to insulin deficiency, can lead to devastating complications. Early detection and intervention are the cornerstones of effective management and improved patient outcomes. This study breaks new ground by meticulously evaluating the performance of each clustering algorithm using advanced metrics like silhouette score and adjusted Rand index. The goal is to identify the method that generates the most accurate and well-defined clusters for diabetes-related attributes. This, in turn, has the potential to revolutionize diabetes diagnosis, enabling earlier interventions and ultimately leading to better disease management and patient care. By providing a comprehensive comparison of these clustering techniques, this research offers a significant contribution to the fight against diabetes
Enhanced deepfake detection using an ensemble of convolutional neural networks
Get PDFDigital media integrity and authenticity have been seriously challenged with the rise of deepfakes. The challenge is to automatically detect this artificial intelligence (AI) generated manipulations. These manipulations or forgeries can cause harmful consequences such as spreading fake news in politics, scamming people online and invading privacy. Convolutional neural networks (CNN) models are found to be good at classification tasks, but the performance could not reach high accuracy, especially when they were tested on more challenging deepfake datasets. In this paper we present a deepfake detection system based on an ensemble of CNN architectures, ResNet50 and EfficientNet, capable of distinguishing between real and deepfake videos with high accuracy. For the experiment, we have chosen Celeb-DF version 2, as it has emerged to be one of the most challenging deepfake dataset. The ensemble model achieved an F1-score of 94.69% and an accuracy of 90.58%, outperforming the individual CNN models. This study shows that ensemble learning can increase the reliability and accuracy of deepfake detection systems on challenging datasets
Imagery based plant disease detection using conventional neural networks and transfer learning
Get PDFEnsuring the sustainability of global food production requires efficient plant disease detection, challenge conventional methods struggle to address promptly. This study explores advanced techniques, including convolutional neural networks (CNNs) and transfer learning models (ResNet and VGG), to improve plant disease identification accuracy. Using a plant disease dataset with 65 classes of healthy and diseased leaves, the research evaluates these models' effectiveness in automating disease recognition. Preprocessing techniques, such as size normalization and data augmentation, are employed to enhance model reliability, and the dataset is divided into training, testing, and validation sets. The CNN model achieved accuracies of 95.45 and 94.52% for 128×128 and 256×256 image sizes, respectively. ResNet50 proved the best performer, reaching 98.38 and 98.63% accuracy, while VGG16 achieved 97.99 and 98.34%. These results highlight ResNet50's superior ability to capture intricate features, making it a robust tool for precision agriculture. This research provides practical solutions for early and accurate disease identification, helping to improve crop management and food security
Integrating IndoBERT and balanced iterative reducing and clustering using hierarchies of BERTopic in Indonesian short text
Get PDFShort text topic modeling remains challenging due to data sparsity, limited word co-occurrences, and unstable clustering results, particularly for Indonesian texts. This study proposes an improved BERTopic framework that integrates IndoBERT embeddings, best match 25 (BM25)-based topic representation, and balanced iterative reducing and clustering using hierarchies (BIRCH) clustering to address these issues. IndoBERT generates contextual embeddings adapted to Indonesian linguistic features, and BM25 weighting improves keyword relevance by considering document length and term saturation. BIRCH clustering minimizes outliers by assigning most documents to valid clusters, which enhances data utilization and topic stability. Experiments on Indonesian datasets from X (formerly Twitter), Google Reviews, and YouTube demonstrate that the proposed approach consistently achieves higher topic coherence. The proposed method yields stable topic diversity values between 0.91 and 0.94, maintains embedding density from 0.60 to 0.66, and achieves intra-topic similarity between 0.39 and 0.41 across increasing dataset sizes. The proposed framework successfully reduces outlier proportions to 1-5%, which significantly outperforms standard BERTopic and K-Means. Furthermore, the model maintains stable topic counts as the data volume grows, confirming robustness and scalability for sparse short text modeling. Overall, integrating IndoBERT, BM25, and BIRCH provides a more coherent, stable, and effective solution for Indonesian short text topic modeling
Urban incident detection based on hybrid convolutional neural networks and bidirectional long short-term memory
Get PDFReal-time incident detection is a major challenge in urban roads. This paper proposes an innovative hybrid method for incident detection, combining convolutional neural networks (CNN) and bidirectional-long short-term memory (Bi-LSTM). CNN extracts complex spatial features from raw data, while Bi-LSTMs are used for incident detection by capturing long-term temporal dependencies present in data. The proposed algorithm is evaluated using simulated data from the open-source software simulation of urban mobility (SUMO). This combination improves incident detection's accuracy and robustness by exploiting spatial and temporal information. Experimental results show that our hybrid approach outperforms the support vector machine (SVM), random forest (RF), and Bi-LSTM algorithms, with a substantial decrease in false positives and the speed of detecting urgent situations