IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Arabic text diacritization using transformers: a comparative study
Get PDFThe Arabic language presents challenges for natural language processing (NLP) tasks. One such challenge is diacritization, which involves adding diacritical marks to Arabic text to enhance readability and disambiguation. Diacritics play a crucial role in determining the correct pronunciation, meaning, and grammatical structure of words and sentences. However, Arabic texts are often written without diacritics, making NLP tasks more complex. This study investigates the efficacy of advanced machine learning models in automatic Arabic text diacritization, with a concentrated focus on the Arabic bidirectional encoder representations from transformers (AraBERT) and bidirectional long short-term memory (Bi-LSTM) models. AraBERT, a bidirectional encoder representation from transformers (BERT) derivative, leverages the transformer architecture to exploit contextual subtleties and discern linguistic patterns within a substantial corpus. Our comprehensive evaluation benchmarks the performance of these models, revealing that AraBERT significantly outperforms the Bi-LSTM with a diacritic error rate (DER) of only 0.81% and an accuracy rate of 98.15%, against the Bi-LSTM's DER of 1.02% and accuracy of 93.88%. The study also explores various optimization strategies to amplify model performance, setting a precedent for future research to enhance Arabic diacritization and contribute to the advancement of Arabic NLP
Smart contracts vulnerabilities detection using ensemble architecture of graphical attention model distillation and inference network
Get PDFSmart contracts are automated agreements executed on a blockchain, offering reliability through their immutable and distributed nature. Yet, their unalterable deployment necessitates precise preemptive security checks, as vulnerabilities could lead to substantial financial damages henceforth testing for vulnerabilities is necessary prior to deployment. This paper presents the graphical attention model distillation and inference network (GAMDI-Net), a pioneering methodology that significantly enhances smart contract vulnerability detection. GAMDI-Net introduces a unique graphical learning module that employs attention mechanism networks to transform complex contract code into a smart graphical representation. In addition to this a dual-modality model distillation and mutual modality learning mechanism, GAMDI-Net excels in synthesizing semantic and control flow data to predict absent bytecode embeddings with high accuracy. This methodology not only improves the precision of vulnerability detection but also addresses scalability and efficiency challenges, reinforcing trust in the deployment of secure smart contracts within the blockchain ecosystem
Enhancing convolutional neural network based model for cheating at online examinations detection
Get PDFIn the last few years, e-learning has revolutioning education, giving students access to diverse and adaptable on-line resources, but it has also face a major challenge: cheating on online exams. Students now use variant cheating methods include consulting unauthorized documents, communicating with others during the exam, searching for information on the internet. Combating these cheating practices has become crucial to preserving the integrity of academic assessments. In this context, artificial intelligence (AI) has emerged as an essential tool for mitigating this fraudulent behavior. Equipped with advanced machine learning capabilities, AI can examine a wide range of data to detect student suspicious behavior. This study develops an approach based on a convolutional neural network (CNN) model designed to detect cheating by analyzing candidates' head movements during online exams. By exploiting the FEI dataset, this model achieves an interesting accuracy of 97.28%. In addition, we compare this model to the well-known transfer learning models used in the literature namely, ResNet50, VGG16, DenseNet21, MobileNetV2, and EfficientNetB0 demonstrating the out performance of our approach in detecting cheating during online exams
Weighted nearest neighbors and radius oversampling for imbalanced data classification
Get PDFThe challenges associated with high-dimensional and imbalanced datasets were observed to often lead to a degradation in the performance of classical machine learning algorithms. In the case of high dimensional data, not all features contribute significantly and are considered relevant to the performance of the model. Therefore, this study introduced a novel method called feature weighted variance analysis-nearest neighbors (WFVANN) which was developed on the foundation of k-nearest neighbors (KNN). The process involved modifying the calculation of the Euclidean distance by fully considering the relevance and contribution levels of features based on their Fvalue. WFVANN at the algorithmic level processing and radius-synthetic minority oversampling technique (R-SMOTE) at the data level processing used as the oversampling method later became the proposed model to solve the aforementioned issues. Moreover, extensive experiments were conducted on two distinct types of data including the high-dimensional and imbalanced by comparing WFVANN with the state-of-art KNN-based and synthetic minority oversampling technique (SMOTE)-based methods. The results showed that the proposed method had the highest accuracy, precision, recall, and F1-measure values across the majority of test datasets and outperformed the other methods
Levenberg-Marquardt-optimized neural network for rainfall forecasting
Get PDFRainfall is a crucial meteorological indicator with applications in agriculture, aviation, and military. Forecasting is essential due to unpredictable environmental changes. Current methods use complex statistical models, which are timeconsuming. The present study is targeted for forecasting rainfall with the help of meteorological parameters, viz., temperature, humidity, wind speed, wind direction, and rain, using a specialized artificial intelligence (AI) method and real-time data captured over the study area. The weather station installed at KLE Dr. M. S. Sheshgiri College of Engineering and Technology in Karnataka, India, collects meteorological data. The models used were principal component regression (PCR) and Levenberg-Marquardt -optimized neural network (LMAONN). The Levenberg-Marquardt (LMA) backpropagation (BP) algorithm performed better than other BP algorithms. The coefficient of determination (R2) observed for the PCR and LMAONN models were 0.57 and 0.87, respectively. The LMAONN model provided a better fit for rainfall forecasting than the PCR model, with an index of agreement (IoA) of 0.96, indicating good forecasting
Advanced methodologies resolving dimensionality complications for autism neuroimaging dataset: a comprehensive guide for beginners
Get PDFAutism spectrum disorder (ASD) is gender biased neurodevelopmental condition consisting of a triad of physiological symptoms. Neural images and neurobiology of cognitive disorders are complex but provide significant information and accurate visualization of developmental changes. The diagnosis is time-consuming and necessitates sufficient evidence to distinguish the disorder from other concomitant diseases. The most recent area of interest for cognitive research is neuroimaging, which is used to study the disorder's impact, affected region, and functional connectivity between the regions of interest. The challenges in the domain are the availability of data, the modalities of data, the selection of the correct processing strategies, and the result assessment complications. The study employed machine learning (ML) methods to process the autism data in both structural and functional data formats collected from the autism brain imaging data exchange (ABIDE) consortium. A comparative analysis among image processing methodologies with both data formats was successfully implemented. The variations in the processing pipeline and the outputs strongly suggest an emerging need for 3D/4D images to visualize better, accurate feature extraction and classification. The study aims to support the researchers in identifying the correct image format for specific objectives and the ML techniques, such as Gaussian median filters, segmentation methodologies for 2D data, or a well-defined preprocessing pipeline for 3D data, to achieve reliable and generalized results
Detecting human fall using internet of things devices for healthcare applications
Get PDFFalls pose a significant threat to unintentional injuries, particularly impacting the independence of older individuals. Existing detection methods suffer from drawbacks, including inaccuracies, wearer discomfort, complex setup, resource-intensive computation, and limitations in detecting falls outside a specific setting. In response, our innovative fall detection system integrates with a pneumatic solution, analyzing fundamental human activities like running, walking, and sitting, both indoors and outdoors. This approach combines wearable sensors with a vision-based solution, utilizing a smart belt with embedded accelerometer and gyroscope, alongside wall-installed cameras in a smart house. The system triggers an airbag and sends an emergency alarm upon fall detection. To achieve this, we propose FallMixer a lightweight deep learning model, combined with ‘you only look once’ version 8 (YOLOv8) algorithm, fine-tuned on a collected video dataset to enable real-time detection. We found that the models result in competitive performance, as demonstrated on SisFall, UCI human activity recognition (HAR), and mobile health (MHEALTH) datasets with a remarkable mean average precision. Subsequently, we assess the hardware performance of our solution on edge devices
Squeeze-excitation half U-Net and synthetic minority oversampling technique oversampling for papilledema image classification
Get PDFThe emergence of various convolutional neural networks (CNN) architectures indicates progress in the computer vision field. However, most of the architectures have large parameters, which tends to increase the computational cost of the training process. Additionaly, imbalanced data sources are often encountered, causing the model to overfit. The aim of this study is to evaluate a new method to classify retinal fundus images from imbalanced data into the corresponding classes by using fewer parameters than the previous method. To achieve this, squeeze-excitation half U-Net (SEHUNET) architecture, a modification of half U-Net with squeeze-excite process to provide attention mechanism on each feature maps channel of the model, in combination with synthetic minority oversampling technique (SMOTE) is proposed. The test accuracy of SEHUNET is 98.52% with area under the curve of receiver operation characteristic (AUROC) of 0.999. This result outperforms the previous study that used CNN with Bayesian optimization, achieving accuracy of 95.89% and AUROC of 0.992. SEHUNET is also able to compete with the transfer learning methods used in previous research such as InceptionV3 with 96.35% accuracy, visual geometry group (VGG) with 96.8%, and ResNet with 98.63%. This performance can be achieved by SEHUNET with only 0.268 million parameters compared to the architecture parameters used in previous research ranging from 11 million to 33 million
Towards efficient knowledge extraction: Natural language processing-based summarization of research paper introductions
Get PDFAcademic and research papers serve as valuable platforms for disseminating expertise and discoveries to diverse audiences. The growing volume of academic papers, with nearly 7 million new publications annually, presents a formidable challenge for students and researchers alike. Consequently, the development of research paper summarization tools has become crucial to distilling crucial insights efficiently. This study examines the effectiveness of pre-trained models like text-to-text transfer transformer (T5), bidirectional encoder representations from transformers (BERT), bidirectional and auto-regressive transformer (BART), and pre-training with extracted gap-sentences for abstractive summarization (PEGASUS) on research papers, introducing a novel hybrid model merging extractive and abstractive techniques. Comparative analysis of summaries, recall-oriented understudy for gisting evaluation (ROUGE) and bilingual evaluation understudy (BLEU) score evaluations and author evaluation help evaluate the quality and accuracy of the generated summaries. This advancement contributes to enhancing the accessibility and efficiency of assimilating complex academic content, emphasizing the importance of advanced summarization tools in promoting the accessibility of academic knowledge
Convolutional neural network modelling for autistic individualized education chatbot
Get PDFThe traditional education system for autistic kids needs integration with computer technology that embraces artificial intelligence to help school instructors and management. An application that enables the teacher to retrieve information from a trusted source is essential since the information is only sometimes available on time. Thus, developing a chatbot application that utilizes natural language processing can enhance the management of autistic schools and will help individualized education for autistic students. This research uses a deep learning model that utilizes a convolutional neural network to develop a chatbot as a teaching assist tool for teachers. The results show that the chatbot has achieved ˜0.03% loss when trained with different epoch numbers. In terms of usability, the chatbot achieves mean system usability scores of 80.48 ± 13.03. This may open opportunities for more effective individualized education for students with special needs and increase the potential to improve inclusive education for disabled students. It is useful to include future actions that enable the simplification of the use of this chatbot tool in a wide range of contexts. To close the education gap for children with disabilities, chatbots could help people with communication disabilities and could also significantly enhance the rate of communication