IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Optimizing bioinformatics applications: a novel approach with human protein data and data mining techniques
Get PDFBiomedicine plays a crucial role in medical research, particularly in optimizing techniques for disease prediction. However, selecting effective optimization methods and managing vast amounts of medical data pose significant challenges. This study introduces a novel optimization technique, integrated bioinformatics optimization model (IBOM) for disease diagnosis, incorporating data mining to efficiently store large datasets for future analysis. Various optimization algorithms, such as whale optimization algorithm (WOA), multi-verse optimization (MVO), genetic algorithm (GA), and ant colony optimization (ACO), were compared with the proposed method. The evaluation focused on metrics like accuracy, specificity, sensitivity, precision, F-score, error, receiver operating characteristic (ROC), and false positive rate (FPR) using 5-fold cross-validation. Results indicated that the 5-fold cross-validation method achieved superior performance with metrics: 98.61% accuracy, 96.59% specificity, 88.63% sensitivity, 99.30% precision, 92.31% F-score, 10.80% error, 92.61% ROC, and a 3.00% FPR. This method was found to be the most effective, achieving an accuracy of 0.92 in disease diagnosis compared to other optimization techniques
Evaluating search key distribution impact on searching performance in large data streams
Get PDFThe distribution pattern of search keys is assessed in this study by contrasting four methods of index searching on large-scale JSON files with data streams. The Adelson-Velskii and Landis (AVL) tree, binary search tree (BST), linear search (LS), and binary search (BS) are among the search strategies. We look at the normal distribution, left-skewed distribution, and right-skewed distribution of search-key distributions. According to the results, LS performs the slowest, averaging 653.166 milliseconds, whereas AVL tree performs better than the others in dense index, with an average search time of 0.005 milliseconds. With 0.011 milliseconds per keyword for sparse index, BS outperforms LS, which averages 1007.848 milliseconds. For dense indexing, an AVL tree works best; for sparse indexing, BS is recommended
Comparative analysis of genetic algorithms for automated test case generation to support software quality
Get PDFSoftware testing is crucial for enhancing software quality, but designing test cases is a labor-intensive, resource-intensive, and time-consuming process. Additionally, test case designers often introduce subjectivity when creating test cases manually. To address these challenges, this paper compares three different approaches for automatically generating program branch coverage test cases: the parallel data generation algorithm (PDGA), a standard genetic algorithm (SGA), and a random test generation method. By leveraging genetic algorithms and parallel data generation techniques, these automated approaches aim to reduce the manual effort, resources, and potential biases involved in test case design, while improving the efficiency and effectiveness of achieving comprehensive branch coverage during software testing. The experimental results, conducted using five datasets with programs written in PHP, demonstrate that PDGA outperforms both SGA and random methods across various tested programs, achieving higher maximum and average coverage. Specifically, PDGA achieved an average coverage of 100% in the "calculator" program, highlighting its superior stability and efficiency. While SGA also shows good performance, it is not as optimal as PDGA, and the random method shows the lowest performance among the three. These findings underscore the potential of genetic algorithms, particularly PDGA, to enhance the coverage and quality of software testing, thereby significantly improving system reliability.
Machine learning methods for classification and prediction information security risk assessment
Get PDFInformation is an essential company asset that must be protected. The value of information assets depends on the type and scale of the business and its role in delivering services. One of the primary programs that can help identify areas of improvement and guide the development of security awareness programs is risk assessment. Managing cybersecurity risks is critical to protecting enterprises from developing cyber threats and promoting resilience. This includes detecting, assessing, and mitigating risks to protect sensitive data, systems, and networks. While cybersecurity risk management is challenging, organizations may improve their security posture. This paper seeks to contribute to the field of information security risk assessment by leveraging the power of machine learning to provide quick, cost-effective, and individualized risk assessments for small and medium enterprises. Specifically, we extend the evaluation for security level classification by utilizing a support vector machine, random forest, and gradient boosting algorithms. The results demonstrate how well the model detects significant cases while reducing false positives. The model’s exceptional precision ensures that its identifications are dependable, while the high recall demonstrates that it accurately detects relevant data. Precision is critical in security risk assessment because a false positive result might have profound effects
A systematic analysis on machine learning classifiers with data pre-processing to detect anti-pattern from source code
Get PDFAutomatic detection of anti-patterns from source code can reduce software maintenance costs massively. Nowadays, machine learning approaches are very commonly used to identify anti-patterns. Hence, it is very crucial to choose a classifier that can be useful for detecting anti-patterns. This work aims to help practitioners to choose a suitable classifier to detect anti-patterns. In this paper, we highlight 16 classifiers in four different categories to detect anti-patterns. Furthermore, the performance of these classifiers is identified with the data pre-processing (DPP) to detect four commonly occurring anti-patterns from the three commonly used open-source Java projects’ source code. The accuracy of Dagging classifiers is 98.4%. Kernel logistic regression (KLR) also performs well i.e., 97%. In the case of time complexity, naive Bayes (NB), decision trees (DT), support vector machines (SVM), library for support vector machines (LibSVM), logistic, and LightGBM (LB) have less time complexity to build a model in all the projects
A Fletcher-Reeves conjugate gradient algorithm-based neuromodel for smart grid stability analysis
Get PDFInterest in smart grid systems is growing around the globe as they are getting increasingly popular for their efficiency and cost reduction at both ends of the energy spectrum. This study, therefore, proposes a neuro model designed and optimized with the Fletcher-Reeves conjugate gradient algorithm for analyzing the stability of smart grids. The performance results achieved with this algorithm was compared with those obtained when the same network was trained with other algorithms. Our results show that the proposed model outperforms existing techniques in terms of accuracy, efficiency, and speed. This study contributes to the development of intelligent solutions for smart grid stability analysis, which can enhance the reliability and sustainability of power systems
Fastest Moroccan license plate recognition using a lightweight modified YOLOv5 model
Get PDFMorocco is witnessing an alarming surge in road accidents. Automatic license plate recognition (ALPR) technology is vital in enhancing road safety. It en- ables applications like traffic management, law enforcement, and toll collection by automatically identifying vehicles on the roads. This paper integrated the ShuffleNet V2 architecture into the end-to-end YOLOv5 object detection sys- tem. The goal was to develop a model capable of accurately detecting Moroc- can license plates with an 87% accuracy rate. The proposed model was able to achieve high processing speeds of 60 frames per second (FPS) while maintain- ing a compact size of 1.3 megabytes and a limited computational requirement of 0.44 million floating-point operations. Compared to other models used in similar contexts, this model demonstrates superior performance and high com- patibility with embedded systems, making it a promising solution for addressing road safety challenges in Morocco
Hyperparameter optimization of convolutional neural network using particle swarm optimization for emotion recognition
Get PDFEmotion identification has been widely researched based on facial expressions, voice, and body movements. Several studies on emotion recognition have also been performed using electroencephalography (EEG) signals and the results also show that the technique has a high level of accuracy. EEG signals that detected by standart method using exclusive representations of time and frequency domains presented unefficient results. Some researchers using the convolutional neural network (CNN) method performed EEG signal for emotional recognition and obtained the best results in almost all benchmarks. Although CNN has shown fairly high accuracy, there is still a lot of room for improvement. CNN is sensitive to its hyperparameter value because it has considerable effect on the behavior and efficiency of the CNN architecture. So that the use of optimization algorithms is expected to provide an alternative selection of appropriate hyper parameter values on CNN. Particle swarm optimization (PSO) algorithm is a metaheuristic-based optimization algorithm with many advantages. This PSO algorithm was chosen to optimize the hyperparameter values on CNN. Based on the evaluation results in each model, hybrid CNN-PSO showed better results and achieved the best value in 80:20 split data which is 99.30% accuracy
Sign language emotion and alphabet recognition with hand gestures using convolution neural network
Get PDFAmerican sign language (ASL) is a special means of interaction for hard-of-hearing individuals and has precise conventional rules. Since the general public does not know these sign language protocols, there is a need to have an efficient automatic sign-emotion recognition system. The objective of this paper is, to develop a framework that recognizes standard hand gestures. The gesture represents emotions and alphabet. This paper covers the methodology, results and performance factors, for experimentations. This experimentation of ASL-based alphabet and emotion recognition is novel as till now many efforts of alphabets categorization are done but this is the new direction of research where emotions, such as together’, ‘happy’, ‘peace, ’sad’, ‘confused’, and ‘love’ are captured and automatically classified with hand signs. We mention our approach to increase ‘accuracy’, wherein we capture images and regions of interest (ROI). In this article, a specifically designed convolution neural network (CNN), is used to identify emotions from hand gestures and the addition of ROI enhances accuracy. The captured hand gesture dataset of the size of 94,000 images. “peace” sign emotion has the highest recognition rate (‘98.95%’). Alphabet’s “P” and “Q” sign ASL alphabets have the maximum recognition rate of signs. In all, very impressive accuracy of “92%” and above is detected. The limits of the experimentation are as mentioned i) there is no repeatability of accuracy for the same hand gesture; ii) The distance and angle of hand gestures with camera are crucial factors for an experiment; and iii) the alphabet recognition system is not working for the alphabets “J” and “Z”
Hybrid model detection and classification of lung cancer
Get PDFLung cancer ranks among the most prevalent malignancies worldwide. Early detection is pivotal to improving treatment outcomes for various cancer types. The integration of artificial intelligence (AI) into image processing, coupled with the availability of comprehensive historical lung cancer datasets, provides the chance to create a classification model based on deep learning, thus improving the precision and effectiveness of detecting lung cancer. This not only aids laboratory teams but also contributes to reducing the time to diagnosis and associated costs. Consequently, early detection serves to conserve resources and, more significantly, human lives. This study proposes convolutional neural network (CNN) models and transfer learning-based architectures, including ResNet50, VGG19, DenseNet169, and InceptionV3, for lung cancer classification. An ensemble approach is used to enhance overall cancer detection performance. The proposed ensemble model, composed of five effective models, achieves an F1-score of 97.77% and an accuracy rate of 97.5% on the IQ-OTH/NCCD test dataset. These findings highlight the effectiveness and dependability of our novel model in automating the classification of lung cancer, outperforming prior research efforts, streamlining diagnosis processes, and ultimately contributing to the preservation of patients' lives