IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Transforming images into words: optical character recognition solutions for image text extraction
Get PDFOptical character recognition (OCR) tool is a boon and greatest advancement in today’s emerging technology which has proven its remarkability in recent years by making it easier for humans to convert the textual information in images or physical documents into text data making it useful for analysis, automation processes and improvised productivity for different purposes. This paper presents the designing, development and implementation of a novel OCR tool aiming at text extraction and recognition tasks. The tool incorporates advanced techniques such as computer vision and natural language processing (NLP) which offer powerful performance for various document types. The performance of the tool is subject to metrics like analysis, accuracy, speed, and document format compatibility. The developed OCR tool provides an accuracy of 98.8% upon execution providing a character error rate of 2.4% and word error rate (WER) of 2.8%. OCR tool finds its applications in document digitization, personal identification, archival of valuable documents, processing of invoices, and other documents. OCR tool holds an immense amount of value for researchers, practitioners and many organizations which seek effective techniques for relevant and accurate text extraction and recognition tasks
Automatic diabetes prediction with explainable machine learning techniques
Get PDFDiabetes is a metabolic disorder caused by various genetic, physiological and behavioral factors. It occurs due to an imbalance in the body’s insulin processing, which results in elevated blood sugar levels. Its early diagnosis can alleviate the risk of other deadly diseases. The onset and accurate detection of diabetes can decrease the progression of different complications and dysfunction of tissues. The principal objective of this article is to utilize machine learning approaches to predict the existence of diabetes in female patients at a primary stage. Multiple machine learning, including ensemble classifiers with the Pima Indian dataset and a private dataset obtained from a local Bangladeshi hospital, are used in this work. We employed feature scaling, synthetic oversampling technique (SMOTE), and hyperparameter optimization with GridSearchCV to get the best performance from different machine learning algorithms. The support vector machine (SVM) with the SMOTE framework and default hyperparameters achieved the accuracy and F1 score of 87% and 91%, respectively. The accuracy and F1 score of the SVM model improved to 95% and 91%, respectively, with hyperparameter optimization. Finally, explainable artificial intelligence with the local interpretable model-agnostic explanations (LIME) is employed to illustrate the predictability of the SVM technique
A contest of sentiment analysis: k-nearest neighbor versus neural network
Get PDFDiscourse about public matters often encompasses sentences that address topics emerging within societal contexts, including issues related to Islamophobia. Debates surrounding this subject frequently evoke support and opposition within digital platforms and interpersonal interactions. Categorizing such dialogic expressions within online media facilitates an evaluation of their negative and positive implications. This study employs two distinct methodologies, specifically deep learning and machine learning techniques, to visualize the findings by implementing dual algorithms. According to the comparative analysis, deep learning achieves a higher accuracy rate of 78%, whereas machine learning achieves a rate of 71%. Thus, deep learning is a better method for textual data classification
Machine learning for mental health: predicting transitions from addiction to illness
Get PDFThe increasing prevalence of infection-causing diseases due to environmental factors and lifestyle choices has strained the healthcare system, necessitating advanced techniques to save lives. Disease prediction plays a crucial role in identifying individuals at risk, enabling early treatment, and benefiting governments and health insurance providers. The collaboration between biomedicine and data science, particularly artificial intelligence and machine learning, has led to significant advancements in this field. However, researchers face challenges related to data availability and quality. Clinical and hospital data, crucial for accurate predictions, are often confidential and not freely accessible. Moreover, healthcare data is predominantly unstructured, requiring extensive cleaning, preprocessing, and labeling. This study aims to predict the likelihood of patients transitioning to mental illness by monitoring addiction conditions and constructing treatment protocols, with the goal of modifying these protocols accordingly. We focus on predicting such transformations to illuminate the underlying factors behind shifts in mental health. To achieve this objective, data from an Iraqi hospital has been collected and analyzed yielding promising results.
A proposed approach for plagiarism detection in Myanmar Unicode text
Get PDFAround the world, with technology that improves over time, almost everyone can access the internet easily and quickly. With the increase in the use of the internet, the plagiarism of information that is easily available on the internet has also increased. Such plagiarism seriously undermines originality and ethical principles. In order to prevent these incidents, there is plagiarism detection software for many countries and languages, but there is no plagiarism detection software for the Myanmar language yet. In an attempt to fill that gap, this study proposed a deep learning model with Rabin-Karp hash code and Word2vec model and built a plagiarism detection system. Our deep learning model was trained by randomly obtaining information from Myanmar Wikipedia. According to the experiments, our proposed model can effectively detect plagiarism of educational content and information from Myanmar Wikipedia. Moreover, it is possible to distinguish plagiarized texts by rearranging words or substituting words with some synonyms. This study contributes to a broader understanding of the complexities of plagiarism in the Myanmar academic area and highlights the importance of measures to effectively prevent plagiarism. It maintains the credibility of education and promotes a culture that values originality and intellectual integrity
Artificial intelligence and machine learning adoption in the financial sector: a holistic review
Get PDFThe evolution of new technologies has spurred a growing body of literature exploring their application and impact on the financial sector, particularly the integration of artificial intelligence (AI). This paper delves into the rapid adoption of AI and machine learning within the financial sector, highlighting their potential to enhance financial stability and productivity. By reviewing research from 2018 to 2023, the study categorizes AI applications in finance into three main areas: cybersecurity, customer services, and financial management. Furthermore, the research identifies and classifies various threats posed to the integrity and stability of the financial system by AI, along with associated challenges for policy and regulatory frameworks. It also addresses the risks and obstacles inherent in deploying AI within financial markets and banking sectors, offering recommended strategies to mitigate these limitations. Despite the recognized advantages, the comprehensive understanding of AI's benefits and drawbacks remains incomplete due to its evolving nature and varied applications in banking. Clear policies governing AI usage are imperative to safeguard financial consumers and promote a fair and transparent financial market. These guidelines should prioritize human decision-making and foster an unbiased approach to policymaking, ultimately fostering innovation within the industry
Enhancing fall detection and classification using Jarratt‐butterfly optimization algorithm with deep learning
Get PDFFalls pose significant risk to the health and safety of individuals, specifically for vulnerable populations as the elderly and those with specific medical conditions. The repercussions of falls can be severe, leading to injuries, loss of independence, and increased healthcare costs. Consequently, the development of effective fall detection systems is crucial for providing timely assistance and enhancing the overall well-being of affected individuals. Recent advancements in deep learning (DL) have opened new avenues for automating fall detection through the analysis of sensor data and video footage. DL algorithms are especially well-suited for this task because they can automatically learn complex features and patterns from raw data, eliminating the need for extensive manual feature engineering. This article introduces a novel approach to fall detection and classification, termed the fall detection and classification using Jarratt‐butterfly optimization algorithm with deep learning (FDC-JBOADL) algorithm. The FDC-JBOADL technique employs a median filtering (MF) method to mitigate noise and utilizes the EfficientNet model for robust feature extraction, capturing both motion patterns and appearance characteristics of individuals. Furthermore, the classification of fall events is achieved through a long short-term memory (LSTM) classifier, with hyperparameter optimization facilitated by Jarratt‐butterfly optimization algorithm (JBOA). Through a comprehensive series of experiments, the efficacy of FDC-JBOADL technique is validated, demonstrating superior performance compared to existing methodologies in the domain of fall detection
Enhancing fire detection capabilities: Leveraging you only look once for swift and accurate prediction
Get PDFDetecting fires is crucial to prevent potentially catastrophic outcomes. Traditional fire detection methods, relying on electronic, chemical, or mechanical sensors, often suffer from time delays in activation due to threshold parameters. An emerging alternative utilizes artificial intelligence, particularly image-based fire detection, using convolutional neural networks (CNNs). You only look once (YOLO) is a state-of-the-art object detection framework prized for speed and real-time capabilities. In our research, we conducted multiple training experiments employing various deep neural network (DNN) architectures as feature extractors for object detection within the YOLOv5 framework. These architectures included MobileNetV3, ResNet, and CSP-Darknet53. Among these configurations, YOLOv5 with CSP-Darknet53 (scale s) emerged as the most accurate, boasting mAP@50 of 0.88 and an impressive FPS of 73, with training model size of 14.50 MB. Furthermore, we integrated the selected model with the streamlit package to create a user-friendly web application interface for fire detection testing. The resulting model demonstrates remarkable efficiency, detecting fires within 0.01 seconds. This research represents a significant advancement in fire detection technology, offering both rapid detection and enhanced accuracy, with potential applications in various settings, from indoor facilities to outdoor environments
Optimizing potato crop productivity: a meteorological analysis and machine learning approach
Get PDFMotivated by the critical need to enhance potato production in Bangladesh, particularly in the face of a changing climate, this study investigates the significant impact of weather on potato yield. This research employs various statistical and machine-learning approaches to identify key weather factors influencing potato crops. We utilize ANOVA F regression and random forest (RF) with feature importance analysis to pinpoint crucial monthly weather variables. Additionally, a correlation study employing Pearson's and Spearman's coefficients alongside p-values is conducted to determine the relationships between weather conditions and crop yield. Seaborn's bivariate kernel density estimation is then used to visualize ideal weather conditions for optimal harvests. Furthermore, to predict future yields, the study implements thoroughly trained and validated machine learning models including k-nearest neighbors (KNN), RF, and support vector regressor (SVR). Our analysis reveals that the RF model emerges as the most reliable predictor, achieving a high correlation coefficient (R²=0.9990), and minimal error values (mean absolute percentage error (MAPE)=0.70, mean absolute error (MAE)=0.0803, and root mean square error (RMSE)=0.1114). These findings provide valuable insights to guide informed agricultural decisions and climate-related strategies, particularly for resource-limited countries like Bangladesh
Imitation of the human upper limb by convolutional neural networks
Get PDFThe paper outlines the development of an algorithm focused on imitating movements of a human arm and replicating strokes generated by the user's hand within a working environment. The algorithm was crafted to discern the position of either the user's left or right arm, tracking each section (fingers, wrist, elbow, and shoulder) through a detection and tracking system. These movements are then replicated onto a virtual arm, simulating the actions of a cutting tool, generating strokes as it moves. Convolutional neural networks (CNNs) were employed to detect and classify each arm section, while geometric analysis determined the rotation angles of each joint, facilitating the virtual robot's motion. The stroke replication program achieved an 84.2% accuracy in stroke execution, gauged by the closure of the polygon, distance between initial and final drawing points, and generated noise, which was under 10%, with a 99% probability of drawing a closed polygon. A Fast region-based convolutional neural network (Fast R-CNN) network detected each arm section with 60.2% accuracy, producing detection boxes with precision ranging from 17% to 59%. Any recognition shortcomings were addressed through mathematical estimation of missing points and noise filters, resulting in a 90.4% imitation rate of human upper limb movement