IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Building change detection via classification in high-resolution aerial imagery
Get PDFThis research investigates the detection of changes in building structures within high-resolution aerial images of Baghdad, Iraq, over two years, 2007 and 2024. Employing advanced remote sensing techniques and sophisticated image processing algorithms, this study aims to identify and quantify alterations in the urban landscape accurately by addressing the key challenges inherent in the image registration process, as well as the availability associated with change detection (CD) techniques. We examined the data collection strategies, evaluated matching methods, and compared CD approaches. Aerial images were accurately analyzed to detect changes in building footprints, construction activities, and destruction. We developed a comprehensive annotation methodology tailored to the complex urban environment of Baghdad. These findings emphasize the rapidly evolving nature of Baghdad’s urban fabric and the critical need for ongoing monitoring to inform urban planning and management strategies. The results demonstrate the efficacy of utilizing high-resolution aerial imagery with object-based CD techniques for detailed urban analysis. This research advances the existing knowledge by providing a robust framework for urban CD, with implications for enhancing urban planning and policy-making processes. Future research will focus on refining the annotation processes and incorporating additional data sources to enhance the accuracy and comprehensiveness of urban CD methodologies
The effectiveness of ChatGPT in extracting architectural patterns and tactics
Get PDFThis work investigates the potential of ChatGPT, a cutting-edge large language model (LLM), for software design analysis specifically in detecting architectural patterns and tactics. The evaluation involves comparing ChatGPT’s performance with that of Archie, a traditional Eclipse plugin designed for architectural analysis. The study uses the source code of five open-source software systems as the testing ground. Results reveal that ChatGPT achieves noteworthy performance in both pattern and tactic detection tasks. Specifically, for pattern detection, ChatGPT demonstrates an accuracy of up to 47.06%, while for tactic detection, it achieves a precision of 28.25%. While ChatGPT’s current capabilities are not yet a replacement for specialized tools like Archie, it offers significant potential as a complementary tool in architectural analysis workflows. By bridging the gap between natural language understanding and software engineering, ChatGPT could pave the way for more intelligent and automated solutions in the field. However, a key limitation is its difficulties in handling foundational or traditional tactics, resulting in a lower detection rate in certain areas. This research contributes valuable insights into the application of LLMs in software engineering, highlighting both the strengths and the limitations of ChatGPT in addressing complex architectural tasks
Residual edge dense enhanced module network: a deep learning approach with multi-class SVM for lung tumor stage classification
Get PDFLung cancer segmentation with positron emission tomography (PET) and computed tomography (CT) images plays a critical role to accurately detect lung cancer. Nevertheless, lung tumor segmentation in PET/CT images were extremely difficult due to the movement caused by respiration. Despite this fact, the lung tumor images shown large number of variations mostly in PET images and CT images. As PET-CT images are acquired concurrently the shape and size of lung tumor varies according to modality. To address these issues, we developed a residual edge dense enhanced module network (REDEM-NET) framework for lung tumor stage classification. The proposed REDEM-NET can process PET and CT images as inputs. In addition, the dense residual convolutional network (DRCN) collects both inputs and extracts high-dimensional features concurrently. The extracted features from both imaging modalities were fed into UNet+++ to obtain multi-level decoded features. The extracted decoded features are concurrently supplied to the pixel level learning module (PELM) and edge level learning module (E2LM) which resulting in two outputs for subsequent learning. The outputs were merged to provide a very precise lung tumor segmentation. Furthermore, segmented tumor was fed to multi-class support vector machine (MC-SVM) for lung tumor stage classification. Moreover, it was able to identify three stages and its substages namely primary tumor, region lymph node and distant metastasis
Enhancing crude palm oil quality detection using machine learning techniques
Get PDFIndonesia, a leading nation in the palm oil industry, experienced a significant increase of 15.62% in crude palm oil (CPO) exports in 2020, effectively meeting the global need for vegetable oil and fat. Therefore, the subjective assessment of CPO quality, influenced by differences in human evaluations, may lead to inconsistencies, necessitating the adoption of machine learning methods. There are several categories of CPO, such as bad and excellent. Machine learning can determine the quality of CPO itself. This study utilizes two distinct categories to measure the quality of CPO. CPO quality data is collected and processed into pre-processing data, in classifying using several methods such as artificial neural network (ANN), k-nearest neighbor (KNN), support vector machine (SVM), decision tree (DT), naïve Bayes (NB), and C.45 using the cross-validation evaluation parameter. The best results are obtained by C.45 and DT with an accuracy of 99.98%
Novel framework for downsizing the massive data in internet of things using artificial intelligence
Get PDFThe increasing demands of large-scale network system towards data acquisition and control from multiple sources has led to the proliferated adoption of internet of things (IoT) that is further witnessed with massive generation of voluminous data. Review of literature showcases the scope and problems associated with data compression approaches towards massive scale of heterogeneous data management in IoT. Therefore, the proposed study addresses this problem by introducing a novel computational framework that is capable of downsizing the data by harnessing the potential problem-solving characteristic of artificial intelligence (AI). The scheme is presented in form of triple-layered architecture considering layer with IoT devices, fog layer, and distributed cloud storage layer. The mechanism of downsizing is carried out using deep learning approach to predict the probability of data to be downsized. The quantified outcome of study shows significant data downsizing performance with higher predictive accuracy
Transformer+transformer architecture for image captioning in Indonesian language
Get PDFImage captioning in Indonesian language poses a significant challenge due to the complex interplay between visual and linguistic comprehension, as well as the scarcity of publicly available datasets. Despite considerable advancements in this field, research specifically targeting the Indonesian language remains scarce. In this paper, we propose a novel image captioning model employing a transformer-based architecture for both the encoder and decoder components. Our model is trained and evaluated on the pre-translated Flickr30k dataset in the Indonesian language. We conduct a comparative analysis of various transformertransformer configurations and convolutional neural network (CNN)-recurrent neural network (RNN) architectures. Our findings highlight the superior performance of a vision transformer (ViT) as the visual encoder, combined with IndoBERT as the textual decoder. This architecture achieved a BLEU-4 score of 0.223 and a ROUGE-L score of 0.472
A machine learning-based approach for detecting communication failures in internet of things networks
Get PDFIn industrial systems, the exchange of massive content, such as high-quality video and large sensing data, among industrial internet of things devices (IIoTDs) is essential, often under strict deadlines. Utilizing millimeter-wave (mmWave) frequencies at 28 and 60 GHz can meet the requirements of industrial internet of things (IIoT) by offering high data rates. However, in the mmWave band, the use of directional antennas is imperative due to the short wavelength, rendering directional links susceptible to adverse effects like deafness problems, where a communicating node fails to receive signals from other transmitting nodes. To mitigate the deafness problem, this paper proposes a machine learning-based communication failure identification scheme for reliable device-to-device (D2D) communication in the mmWave band. The proposed scheme determines the type of network failure (deafness/interference) based on the IIoTD's state parameters. Furthermore, we introduce machine learning based directional medium access control (ML-DMAC) to enhance throughput and minimize the duration of deafness in D2D communication. Performance evaluations demonstrate that the proposed ML-DMAC outperforms existing schemes, achieving approximately 31% higher aggregate throughput and an 88% reduction in deafness duration
Enhanced Arabic-language cyberbullying detection: deep embedding and transformer (BERT) approaches
Get PDFRecent technological advances in smartphones and communications, including the growth of such online platforms as massive social media networks such as X (formerly known as Twitter) endangers young people and their emotional well-being by exposing them to cyberbullying, taunting, and bullying content. Most proposed approaches for automatically detecting cyberbullying have been developed around the English language, and methods for detecting Arabic-language cyberbullying are scarce. Methods for detecting Arabic-language cyberbullying are especially scarce. This paper aims to enhance the effectiveness of methods for detecting cyberbullying in Arabic-language content. We assembled a dataset of 10,662 X posts, pre-processed the data, and used the kappa tool to verify and enhance the quality of our annotations. We conducted four experiments to test numerous deep learning models for automatically detecting Arabic-language cyberbullying. We first tested a long short-term memory (LSTM) model and a bidirectional long short-term memory (Bi-LSTM) model with several experimental word embeddings. We also tested the LSTM and Bi-LSTM models with a novel pre-trained bidirectional encoder from representations (BERT) and then tested them on a different experimental models BERT again. LSTM-BERT and Bi-LSTM-BERT demonstrated a 97% accuracy. Bi-LSTM with FastText embedding word performed even better, achieving 98% accuracy. As a result, the outcomes are generalized
Electrocardiogram sequences data analytics and classification using unsupervised and supervised machine learning algorithms
Get PDFThis paper explores the prediction of cardiovascular disease (CVD) through the classification of electrocardiogram (ECG) sequences using both supervised and unsupervised machine learning (ML) algorithms. ECG 5000 dataset is considered to perform essential data analytics, clustering, and classification, effectively categorizing ECG heartbeats into optimal groups to forecast CVD. The Elbow and Silhouette methods are applied to estimate optimal number of clusters within the dataset. Using K-means and hierarchical clustering algorithms, the data is grouped into two and five distinguishable clusters, with performance metrics indicating that two clusters are more viable. Subsequently, multiple supervised ML classifiers—including kernel classifiers, support vector machine (SVM), naïve Bayes (NB), decision trees (DT), k-nearest neighbor (KNN) and neural networks (NN)—are trained on the labeled and clustered datasets to ensure accurate classification of ECG sequences and anomaly detection. A novel modified ML classifier, kernel-SVM with Chi-Square (χ²) feature selection, is introduced and demonstrates exceptional performance, achieving an impressive accuracy of 0.9848, recall of 0.9973, and a training time of 1.6944 seconds, surpassing benchmarks from prior research. The results and discussion section includes a comparison of various algorithm performances, affirming that the proposed approach is an alternative to the complex deep learning (DL) and transformer-based models
Lung cancer patients survival prediction using outlier detection and optimized XGBoost
Get PDFThis research aims to improve the prediction’s model for survival time of lung cancer patients by using outlier detection, hyper-parameter optimization, and machine learning technique. The research compares the performance of several methods including multilayer perceptron (MLP), decision tree (DT), linear regression (LR), Bagging, XGBoost, and random forest (RF). The dataset used for the experiment is obtained from the surveillance, epidemiology, and end result (SEER) cancer database, which contains diagnoses data from 2004 to 2015. The total number of records used is 196,031 with 22 features. 10-fold cross-validation is used for training and testing sets. The evaluation metrics are root mean square error (RMSE), mean squared error (MSE), R-squared (R2), and mean absolute error (MAE). The results show that the lung cancer patient survival prediction model using the optimized XGBoost (O-XGBoost) model performs the best with an RMSE of 13.74 and outperforms the baseline-XGBoost model as well as other models. This research will be useful for developing a clinical decision support system for the care of lung cancer patients. Physicians can use the developed model to assess the patient’s chance of survival in order to plan more effective treatment