IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Developing a website for English-speaking practice to English as a foreign language learners at the university level
Get PDFThis study explored the adaptation of the ADDIE instructional model in designing and developing a website for the speaking practice of EFL students at the university level. The feasibility of the website was measured through the evaluation of independent experts from three aspects of rating: web design, instructional content, and language usage. Six lecturers and 64 EFL students were invited to evaluate the website. Two lecturers have expertise in multimedia and informatics, while the four others are two experts in instructional content of English teaching and two lecturers in English linguistic expertise. The assessments exposed that the web is easy to use by students and very practical in supporting students for learning; the content of learning material in the website has manifested the syllabus of English-speaking skill on the specified level; and the language used by the website is matched with the level of students’ language proficiency. Therefore, this study successfully developed a prototype of a web-based language learning product that helps students practice English speaking at the intermediate level
Artificial intelligence algorithms to predict customer satisfaction: a comparative study
Get PDFCustomer satisfaction is the key for every business successful. Therefore, keeping the current customer portfolio and expanding it over time is the main goal for any business. Hence, we need first to satisfy these clients. The customer satisfaction helps to retain consumers of its products, increase the life value of the customer, also make known its brand through positive word of mouth to get a better reputation and thus increase turnover. For this reason, several studies have been conducted on this subject to explore all tools and technologies that will help retain customers and reduce their churn rate. Based on various customer satisfaction studies for different types of businesses, this paper shows the review of promising research areas and artificial intelligence (AI) application models in predicting customer satisfaction. The results of this study allowed the identification of the best algorithms with the highest score of performance metrics that can be applied as part of the customer satisfaction prediction, through a detailed benchmark performed. The result shows that random forest (RF) and gradient boost (GB) algorithms in machine learning (ML) and convolutional neural network - long short-term memory (CNN-LSTM) in deep learning (DL) are giving the best performance. The most used metrics are accuracy andF1-score. In addition, DL models outperform ML models in most cases
Accuracy of neural networks in brain wave diagnosis of schizophrenia
Get PDFThis research explores the application of a modified deep learning model for electroencephalography (EEG) signal classification in the context of schizophrenia diagnosis. This study aims to utilize the temporal and spatial characteristics of EEG data to improve classification accuracy. Four popular convolutional neural network (CNN) architectures, namely LeNet-5, AlexNet, VGG16, and ResNet-18, are adapted to handle 1D EEG signals. In addition, a hybrid architecture of CNN-gated recurrent unit (GRU) and CNN-long short-term memory (LSTM) is proposed to capture spatial and temporal dynamics. The model was evaluated on a dataset consisting of EEG recordings from 14 patients with paranoid schizophrenia and 14 healthy controls. The results show high accuracy and F1 scores for all modified models, with CNN-LSTM and CNN-GRU achieving the highest performance with scores of 0.96 and 0.97, respectively. Receiver operating characteristic (ROC) curves demonstrate the model's ability to distinguish between healthy controls and schizophrenia patients. The proposed model offers a promising approach for automated schizophrenia diagnosis based on EEG signals, potentially assisting clinicians in early detection and intervention. Future work will focus on larger data sets and explore transfer learning techniques to improve the generalization ability of the model
Detecting road damage utilizing retinanet and mobilenet models on edge devices
Get PDFA particular form of road digitalization produces a system that detects road damage automatically and in real time, employing the device to detect road damage as an edge device. The application of RetinaNet152 and MobileNetV2 models for road damage detection on edge devices necessitates a trade-off between high system performance and efficiency. Currently, edge devices have limited storage. In this paper, we explore how tuning hyperparameters with batch size and several optimizers improves system performance on RetinaNet152 and MobileNet models, as well as how they are implemented on edge devices. After tuning hyperparameters in the batch size of the optimizer, the Adam optimizer displayed enhanced performance with mean average precision (mAP), average recall (AR), and F1-score. This implies a positive impact on overall model performance. The MobileNetV2 model's hyperparameter tuning technique significantly improves performance, resulting in faster inference times and overall system performance. This demonstrates that the MobileNetV2 model could be used directly on edge devices to identify road damage. However, the RetinaNet152 model has a lower inference time, which cannot be deployed directly to edge devices. The RetinaNet152 model can be deployed on edge devices; however, a technique for speeding up inference time is essential
Classification of Bharatanatyam postures using tailored features and artificial neural network
Get PDFBharatanatyam is a classical dance form of India that upholds the rich culture of India. This dance is learned under the supervision of Guru, the teacher traditionally called in India. The scarcity of experts resulted in the decline of people practicing this dance. There is a need for leveraging technology in preserving and promoting this traditional dance and propagating it amongst the youth. In this research, it is attempted to develop a methodology for automated classification of Bharatanatyam dance postures. The methodology involves extraction of existing features such as speeded up robust features (SURF) and histogram of oriented gradients (HOG), which are used to train and test an artificial neural network (ANN). The results are corroborated with deep learning architectures such as AlexNet and GoogleNet. The proposed methodology has yielded a classification accuracy of 99.85% as compared with 93.10% and 94.25% of AlexNet and GoogleNet respectively. The proposed method finds applications such as assistance to Bharatanatyam dance teachers, e-learning of dance, and evaluating the correctness of the postures
Machine learning-assisted decision support in industrial manufacturing: a case study on injection molding machine selection
Get PDFSelecting the right injection molding machine for new products remains a challenging task that significantly influences the profitability and flexibility of companies. The conventional approach involves performing theoretical calculations for clamping force, conducting mechanical validations of the mold, and carrying out real trials for new parts. This approach is time-consuming, costly, and requires a high level of expertise to ensure the optimal machine choice. This study explores the use of machine learning (ML) methods for efficient machine selection based on product, material, and mold criteria. Six supervised learning techniques were tested on a dataset comprising 70 plastic parts and five machines. Evaluation metrics like F1-score, recall, precision, and accuracy were used to compare models. The results indicate that ML can provide guidance for predicting machine selection, with a preference for the random forest (RF), decision tree (DT), and support vector machine (SVM) models. The most favorable outcome is demonstrated by the RF model, displaying an accuracy of 93%. In this manner, these findings may be helpful for injection molding businesses that are considering the significance of using classification algorithms in their manufacturing process.
A hybrid feature selection with data-driven approach for cardiovascular disease prediction using machine learning
Get PDFAffecting various disorders of heart and blood vessels mainly cardiovascular diseases (CVDs) is the leading cause of human mortality on the planet. A number of machine learning (ML) based supervised learning approaches existing in the literature have been found useful in the clinical decision support system (CDSS) for detecting CVDs automatically. The challenge, however, is that their performance tends to decline unless the training data is of a certain standard. Several approaches to solving this problem are known as feature selection techniques. Despite several notable advancements in the CVD modeling literature, a weak compendium of research exists in an area which supports the integration of the feature selection approach as a means of enhancing the training quality and thus the prediction accuracy. Against this background, in this paper, we proposed a framework called the cardiovascular disease prediction framework (CVDPF) that integrates ML methods. To support this, we designed and proposed a new hybrid feature selection (HFS) algorithm that aims to reduce the number of parameters. This algorithm adopts several filter methods in order to enhance its performance for the task of feature selection. To improve the prediction accuracy of CVDs, a number of ML tools using the HFS approach has been designed and is termed as machine learning based cardiovascular disease prediction (ML-CVDP). The validation of the framework and the algorithms discussed has been done on the basis of a CVD dataset. The experimental findings demonstrated that CVDPF in combination with HFS outperforms other methods of feature selection available
Enhancing traffic flow through multi-agent reinforcement learning for adaptive traffic light duration control
Get PDFThis study addresses urban traffic congestion through deep learning for traffic signal control (TSC). In contrast to previous research on single traffic light controllers, our approach is tailored to the TSC challenge within a network of two intersections. Employing convolutional neural networks (CNN) in a deep Q-network (DQN) model, our method adopts centralized training and distributed execution (CTDE) within a multi-agent reinforcement learning (MARL) framework. The primary aim is to optimize traffic flow in a twointersection setting, comparing outcomes with baseline strategies. Overcoming scalability and partial observability challenges, our approach demonstrates the efficacy of the CTDE-based MARL framework. Experiments using urban mobility simulation (SUMO) exhibit a 68% performance enhancement over basic traffic light control systems, validating our solution across diverse scenarios. While the study focuses on two intersections, it hints at broader applications in complex settings, presenting a promising avenue for mitigating urban traffic congestion. The research underscores the importance of collaboration within MARL frameworks, contributing significantly to the advancement of adaptive traffic signal control (ATSC) in urban environments for sustainable transportation solutions
DriveNet: A deep learning framework with attention mechanism for early driving maneuver prediction
Get PDFInappropriate driving maneuvers are the leading cause of many car accidents. These accidents can be prevented if they are identified in advance and the driver is given the necessary assistance. Anticipating maneuvers is crucial for driving assistance systems in order to alert drivers and take appropriate measures to avoid or mitigate danger. In this paper, we introduce DriveNet a new approach that combines information about the driver’s behavior as well as the driving environment to predict the driving maneuvers. DriveNet utilizes a combination of convolutional neural network (CNN) and long short-term memory (LSTM) with attention mechanism to extract spatial information and capture long temporal dependencies. We evaluate DriveNet by performing a series of experiments using the publicly available Brain4Cars dataset. The findings show that the proposed approach achieves state-of-the-art performance and outperforms most previous methods. DriveNet has achieved an accuracy of 91.24%, a precision of 90.13%, and a recall of 91.44% for anticipation 4 seconds before the maneuvers occur
New family of error-correcting codes based on genetic algorithms
Get PDFThis paper introduces a novel error-correcting code (ECC) construction and decoding approach utilizing genetic algorithms (GAs). Classical ECCs often struggle with efficiency in correcting multiple errors due to time-consuming matrix-based encoding and decoding processes. Our GA-based method optimizes generator vectors to maximize the minimum distance between codewords, enhancing error correction capabilities. Specifically, we construct a new family of ECCs with code length 31, dimension 12, and minimum distance 7, reducing complexity from O(kn) to O(k(n−k)) by encoding message blocks with vectors instead of matrices. In the decoding phase, the GA effectively corrects errors in received codewords. Experimental results show that at a signal-to-noise ratio (SNR) of 7.7 dB, our method achieves a bit error rate (BER) of 10−5 after only 9 generations of the GA. These results demonstrate improved error correction and decoding performance compared to traditional methods. This study contributes an innovative approach using GAs for error correction, offering simpler encoding and robust performance in coding schemes