VFAST - Virtual Foundation for Advancement of Science and Technology (Pakistan)
Not a member yet
1255 research outputs found
Sort by
Optimizing Machine Learning Classifiers for Credit Card Fraud Detection on Highly Imbalanced Datasets Using PCA and SMOTE Techniques
Get PDFCard fraud detection refers to the process of identifying unauthorized or suspicious transactions made using credit or debit cards. It employs machine learning models, rule-based systems, and anomaly detection techniques to detect patterns indicating potential fraud. There is a growing need for systems that can accurately predict and prevent fraudulent transactions. Reducing financial loss by Implementing advanced detection models to safeguard it from fraud or malicious transactions. Therefore, we proposed machine learning models that will predict credit card fraud at an early stage. Also, the study used feature scaling, Principal Component Analysis (PCA), and the Synthetic Minority Over-sampling Technique (SMOTE) to deal with the class imbalance on the dataset. Moreover, SMOTE is applied to balance the classes by synthesizing examples of the minority class, making classifiers more robust. The results show that LR, SVM, KNN, and XGBoost models correctly predict 97\% of fraudulent and non-fraudulent cases. The Decision Tree and the Random Forest models are capable of achieving at least 96\%, respectively. This research combines advanced machine learning methodologies with real-time processing to give insights into predictive analytics in financial fraud detection, which may enhance accuracy and efficiency in financial security systems
Efficient and Secure Data Aggregation Scheme for Wireless Body Area Network
Get PDFThe growth of Wireless Body Area Networks (WBANs) has become an integral part of electronic healthcare systems. WBANs use sensors to continuously collect patient data for health monitoring, which is later transmitted to a remote location, such as a medical server or mobile device, for record-keeping. Due to the limitations of sensors in terms of energy, power, and computational capacity, efficient data aggregation schemes are required to minimize communication overhead during data transmission in WBANs. Additionally, WBANs necessitate secure data transmission, as the records are private and confidential. In this research, we propose a secure and efficient data aggregation scheme for WBANs that utilizes a variance function for aggregation, an authentication code for secure data access, and ChaCha20 for data encryption
Automated Detection and Recognition of Seven-Segment Digits from Electric Meters Utilizing Digital Image Processing and Machine Learning
Get PDFWith the rise of computers and their importance in various fields, traditional electric meters in Pakistan have been replaced by digital electric meters. These digital meters are more accurate and easier to read as they display readings using Seven-Segment Digits. Currently, for billing purposes, human meter readers manually capture images of these meters using cameras or smartphones, and the readings are then recorded manually. This process is time-consuming and prone to errors due to human involvement.Automating the reading of meter images can significantly improve the accuracy and efficiency of the billing process. However, this task is challenging because the captured images can vary in quality, scale, orientation, lighting conditions, and other factors.To address these challenges, we experimented with different machine learning and deep learning models to automatically recognize meter readings from captured images. Five models were trained and evaluated: K-Nearest Neighbors (KNN), Decision Tree, Support Vector Machine (SVM), Random Forest, and Convolutional Neural Network (CNN). These models were tested on real meter data for digit recognition, achieving an accuracy of up to 98%. This promising result demonstrates the potential for fully automating the meter reading process in the future
Topic Modeling of Quranic Verses using Latent Dirichlet Allocation with English Language: Topic Modeling using LDA
No full textThis study aims to assess the effectiveness of topic modeling in the English translation of the Holy Quran. Topic modeling is a popular text mining technique for uncovering latent semantic patterns in the collection of textual documents and helps to annotate the documents based on these topics. This study identifies the most significant topics in each document as well as grasping an understanding of the topic distribution throughout the document sets. Different steps are performed to acquire the dominant topics in each document and identify the distribution of topics across documents. In this context, the present research work chose to employ Latent Dirichlet Allocation as an unsupervised approach for topic modeling since there is no requirement for a training phase as hidden topics can be discovered throughout the topic modeling process. For this, the word cloud is generated to understand and interpret the results after pre-processing. A dictionary and corpus are created to extract the features from the dataset using the Bag of Words approach. The results are evaluated by calculating the perplexity and coherence score, where high coherence indicates the goodness of well-structured topic models and low perplexity score indicates the correctness of prediction made by the topic models. Lastly, the visualization step is performed
Protection Issues and Challenges within the Cloud: A Survey
Get PDFA vast range of computing services are accessible over the Internet utilizing the cloud network. In recent years the cloud storage sector and several others have integrated and encouraged scholars to explore emerging technology. Because of its infrastructure and computing capabilities, its device, data, and resources are transferred to the Cloud Storage System. No expensive network infrastructure is required for consumers and their service costs are reduced to a minimum. Regardless of its advantages, the transition of local computing into remote computing has provided customers and vendors with a range of protection issues and threats. The reliable third party provides several cloud platforms that present new security risks. The cloud provider offers its services over the Internet and uses numerous web technologies that create new security challenges. This paper discusses the fundamental features, safety issues, threats, and solutions for cloud information. The paper also discusses many open-ended cloud security research issues. Over the past five years, substantial technological innovations have emerged, which can add extra comfort to daily lives at the corporate and private levels. Both the personal and the shared divisions have significantly progressed in the advancement of cloud storage technologies. It was evident recently that many companies and organizations have transferred the rest of their burdens onto the cloud. However, the protection problem is a significant debate regarding centralized IT control, which is web-based and susceptible to different forms of attacks. While security measures are in place throughout the entire era, security continues to be a challenge. Here we performed a report on distributed computation and investigated various kinds of attacks and dangers on this modern invention, including various kinds of attacks and dangers on the latest concept, combined with protection measures and current structures
PDDNet: Deep Learning Based Dental Disease Classification through Panoramic Radiograph Images
Get PDFThe high prevalence of dental cavities is a global public health concern. If untreated, cavities can lead to tooth loss, but timely detection and treatment can prevent this outcome. X-ray imaging provides crucial insights into the structure of teeth and surrounding tissues, enabling dentists to identify issues that may not be immediately visible. However, manual assessment of dental X-rays is time-consuming and prone to errors due to variations in dental structures and limited expertise. Automated analysis technology can reduce dentists’ workload and improve diagnostic accuracy. This study proposes the Prediction of Dental Disease Network (PDDNet), a CNN-based model for classifying three categories of dental disease: cavities, fillings, and implants, using X-ray images. PDDNet’s performance is compared with six well-known deep CNN classifiers: DenseNet-201, Xception, ResNet50V2, Inception-V3, Vgg-19, and EfficientNet-B0. To ensure balanced class distribution and enhance classification accuracy, the ADASYN oversampling technique is employed. PDDNet achieves an impressive accuracy of 99.19%, recall of 99.19%, precision of 99.19%, AUC of 99.97%, and F1-score of 99.17%, outperforming the other classifiers across multiple performance metrics. These findings demonstrate PDDNet’s potential to provide significant assistance to dental professionals in diagnosing dental diseases
Improving Efficiency of Finite Population Distribution Function Using Twofold Auxiliary Information Based on Simple Random Sampling
Get PDFThe main purpose of this article has been suggested two modified ratio type estimators for estimation of finite population distribution function using auxiliary information under simple random sampling. Mathematical expressions such as Mean Square Error (MSE) and Bias are examined up to the first order approximation for all considered estimators in this article. For the ideal value of the kappa constant (K), the minimum MSE value for the recommended estimators has been found. Five actual data sets have been carried out to check the precision of the suggested estimators. The suggested estimators found to be superior and more efficient than the existing estimators for population distribution function. The theoretical and empirical comparisons are also conducted. Hypothetically, the proposed estimators perform better than the existing estimators
Application of Machine Learning Techniques for Predicting Stroke Disease
Get PDFStroke is a cerebrovascular illness caused by a sudden halt in blood flow to the brain, resulting in neurological impairment. Stroke is a major public health problem worldwide, affecting millions of people. It is a significant source of illness and mortality, imposing a significant socio-economic burden. A thorough awareness of the current global situation is required for effective treatments and preventive actions. This research compares data mining techniques for the prediction of stroke illness. Using a dataset obtained from Mayo Hospital, Lahore, that had 2326 instances, each with 11 attributes, we compared the performance of Support Vector Machine (SVM), Random Forest, Neural Network, and K-Nearest Neighbors (KNN) approaches. Orange Data Mining Software was applied to evaluate the data and execute machine learning techniques. The results show that Naïve Bayes is the best method for predicting the prevalence of Stroke disease. The proposed model demonstrates an Area Under the Curve (AUC) of 88.3 %, an accuracy of 80.8%, and notable metrics including an F1-Score and precision.
Bit Pattern based Sindhi Character Recognition using Neural Network
Get PDFIn this paper, bit pattern based character recognition for Sindhi language has been presented. The characters of sindhi language are very much complexed to recognize for particular domain. Although there are many studies that have already been done in this recognition but all those are based on image recognition, to give novelty in the idea our system uses bit patterns for characters and provide outcome on the basis of that input pattern. A data set with nine no. of inputs and six outputs for each character is created. We have used patterns due to the computational complexity constant that are 3X3 matrix for input patterns that are uniquely created for all characters and output will be generated in form of binary pattern for the particular character sequence numbers. This system reads the 3X3 matrix in clock wise pattern to get input pattern and match it to created data set. To train the data we have used a Neural Network Model, Multi-Layer Perceptron (MLP) with significant number of hidden layers to get measurable results. The accuracy of 82.6% has been achieved by the experiment
A Convolutional Neural Network (CNN) Based Framework for Enhanced Diagnosis and Classification of COVID-19 Pneumonia
Get PDFCOVID-19 pneumonia is a persistent worldwide health problem that usually affects the most vulnerable groups in society: the newborn and aged populations. Most of the current endeavors toward handling diagnosis and classification of pneumonia have used numerous techniques for machine learning and deep learning, with a particular focus on COVID-19 pneumonia. However, most of these techniques have raised concerns with regard to diagnostic precision as a result of the limited application of advanced algorithms, datasets whose validation is mostly inadequate and predictive capability. To address these limitations, our research introduces a deep learning-based approach by Convolutional Neural Networks (CNNs), which enhances the performance in classifying COVID-19 pneumonia. Salient features of the proposed method include a four-step process: first, data acquisition from a comprehensive chest X-ray dataset on GitHub; second, processing and analyzing the data through visual means like histograms and scatter plots; third, using CNNs supplemented with techniques for data augmentation in supervised learning; lastly, performance evaluation to benchmark against existing models. The present study uses features from X-ray images with the help of CNN\u27s advanced pattern recognition capabilities in pursuit of achieving better generalization and precision in classification. The model achieved an accuracy of 85.70\% and precision of 88.6%, which surpasses the existing techniques and thereby built the promise of improving the accuracy of the diagnostic process, hence, leading to improved care for the patients, and more so forms the foundation on which future AI-powered healthcare diagnostics are expected to take off