Lahore Garrison University Research Journal of Computer Science and Information Technology
Not a member yet
227 research outputs found
Sort by
Enhancing Cyber Security: A Holistic Strategy for Advanced Malicious Website Prediction Using AdaBoost Algorithm
No full textThe rise of malicious websites poses significant challenges to computer systems, demanding robust methodologies for detection and prediction systems to safeguard users. This study introduces a comprehensive four-step methodology for advanced malicious website prediction, aiming to fortify cyber security measures. The methodology using the AdaBoost algorithm and encompasses data preprocessing and feature extraction stages, utilizing a Kaggle-derived dataset and Python libraries. Performance evaluation demonstrates the exceptional accuracy of the AdaBoost model, reaching 78.45%. Comparative analysis underscores the effectiveness of the approach in predicting malicious websites, providing valuable insights for both future research endeavors and industry practices. Furthermore, the study highlights the potential for cost reduction and seamless integration into existing infrastructure. By offering empirical evidence, this research contributes to the enhancement of cyber security strategies and advocates for further exploration of AdaBoost capabilities and feature selection techniques
Performance Benchmarking of Deep Learning Models for Skin Lesion Classification in Resource-Constrained Environments
Full text linkSkin cancer stands as the most prevalent cancer worldwide because health professionals detect new instances regularly in millions annually. The high curability of skin cancer becomes less effective when patients receive their diagnosis too late, which results in elevated mortality numbers worldwide. The research examines deep learning algorithm deployment for improved dermatological examination via skin cancer identification systems. The study evaluates how Convolutional Neural Networks (CNNs), particularly ResNet-50 and SqueezeNet, maintain diagnostic accuracy by analyzing medical images. The assessment uses diverse skin datasets that include various symptoms of cancer alongside different skin types. The study describes CNN design principles and explains data optimization strategies and augmentation approaches that boost system execution. This paper provides a comparative examination of deep learning models that includes an investigation of their interpretability to determine which features generate precise detection results. The study demonstrates that deep learning presents strong potential to become a diagnostic tool that aids physicians in early skin cancer detection and helps them make better clinical choices. The research contributes significant knowledge for improving medical image analysis techniques and technological progress in dermatology
Review of Flutter Application Safeguards Against Security Breaches
Full text linkMobile applications now perform better and give smoother experience to users because of the increasing adoption of flutter framework in development. However, the security concerns about applications are also increasing with the popularity of flutter. There are important security measures which include finding unusual behavior during use and checking if the code has been changed. Flutter is built to be secured but the real security of an app mostly depends on whether the developer writes safe code and follows good development practices. This study not only considers the security of Flutter apps but it also points out common weaknesses, and suggests ways to fix them. It highlights the importance of safe coding practices and gives clear tips to protect sensitive apps like e-commerce apps made with Flutter. Through a combination of empirical analysis and case-based insights, the paper evaluates existing cybersecurity frameworks across multiple technical layers. The goal is to equip developers and decision-makers with actionable recommendations that enhance security posture and ensure compliance with modern security standards in Flutter-based application ecosystems
A HYBRID DEEP LEARNING MODEL FOR ACCURATE BITCOIN PRICE FORECASTING
No full textThe highly stochastic, nonlinear, and volatile nature of Bitcoin prices poses significant challenges for accurate forecasting using traditional statistical models. To address this, we propose a hybrid deep learning architecture that combines the strengths of Convolutional Neural Networks (CNNs) for spatial feature extraction with Recurrent Neural Networks (RNNs), specifically Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU), for capturing long-term temporal dependencies. This integrated framework effectively models both spatial and temporal patterns from historical Bitcoin price data. The model was trained and evaluated using real-world Bitcoin datasets.Experimental results demonstrate that the proposed CNN+LSTM model outperforms traditional machine learning and standalone deep learning approaches. Specifically, it achieves a Root Mean Square Error (RMSE) of 245.76, a Mean Absolute Error (MAE) of 11.45, a Mean Absolute Percentage Error (MAPE) of 15.68%, an R² score of 0.92, and a Mean Bias Error (MBE) of 6.94. These results highlight the effectiveness and reliability of the proposed hybrid model in enhancing the accuracy and stability of financial time series forecasting, providing valuable insights for traders, investors, and financial analysts
An Intelligent Algorithm for Automatic Runtime Selection of Scheduling Algorithm using Pattern Recognition Techniques
No full textThis paper presents a dynamic central processing unit (CPU) scheduling algorithm selection system that utilizes machine learning to optimize the process execution within a computer system. This study evaluated six scheduling algorithms based on key performance indicators such as CPU utilization, turnaround time, waiting time, response time, and throughput. To overcome the limitations of traditional approaches, the proposed methodology integrates a neural network to identify patterns and dynamically choose the most appropriate algorithm at runtime. The paper reviews related literature, highlighting efforts to enhance scheduling algorithms through machine learning and improvements to conventional methods. The methodology includes developing a process database, applying scheduling algorithms, optimizing the outcomes, and training a neural network for dynamic algorithm selection. The results indicate that the proposed approach outperforms existing algorithms regarding waiting time, turnaround time, throughput, and execution efficiency. Additionally, the methodology offers adaptability to diverse process parameters. The study concludes by underscoring the potential for future advancements, including improvements in algorithm precision, incorporation of additional process characteristics, exploration of advanced pattern recognition techniques, and integration of security measures for real-world applications such as cloud computing, edge computing, and IoT (Internet of things) environments
An ANALYTICAL STUDY OF CELLULAR NETWORKS LTE-U AND WI-FI IN 5G ENVIRONMENT
No full textMobile carriers have begun restructuring their set-ups to discharge their stream of traffic flow in unlicensed bands as a result of sharp growth in mobile traffic. Utilizing the radio resources for operators' transmissions, the new 3GPP technologies LAA and LTE-U operate over the 5 GHz unlicensed spectrum using an unlicensed radio interface. Spectrum expansion via different radio interfaces, including LAA and LTE-U, is essential for 5G. To preserve a fair portion of the unlicensed spectrum, Wi-Fi performance is significantly impacted by the physical and MAC layer schemes for LTE-based systems. The coexistence technique suggested in this article controls back off for LTE-U and Wi-Fi when they reach the 5 GHz bands. The new method distributes a set number of packets to both technologies through a new interface located in a heterogeneous cloud RAN. Future fifth-generation (5G) cellular networks will be designed entirely differently as a result of new research paths. The five technologies covered in this article as device-centric designs, massive MIMO, smarter devices, millimeter waves, and native support for machine-to-machine communications could result in disruptive design modifications to both the architecture and the components. Each technology's essential concepts are presented, along with how they might affect 5G and the lingering research problems
APPLYING VGG-16 FOR CLASSIFICATION OF RICE VARIETIES INCORPORATING COLOR AND TEXTURE FEATURES
Full text linkEnsuring rice variety is crucial for meeting client expectations and upholding high-quality standards. However, developing cost-effective and speedy techniques for evaluating rice quality remains a challenge. According to the variation in demand, rice is available in different grades at different prices and rendering to consumer preferences. Deep learning has promising results in various fields of life as their applications are in agriculture. This present study utilized a deep learning based convolutional neural network (CNN) model VGG-16 to identify and classify rice varieties accurately. The applied model is customized, and seven fully connected layers (FCLs) are added on pre-trained VGG-16 as the number of rice classes is seven namely Kachi, Kachi kainat, Seela, Sufaid, super, Ari, and 1508. The most popular libraries namely TensorFlow and Keras are used to develop the proposed neural network model. The steps of this study are image data acquisition using the camera, pre-processing, using the Keras image data generator for transforming data, defining layers, model compilation, training and validation of the model, model optimization and testing on new instances for reliability and accuracy check. The methodology has been evaluated in terms of loss, accuracy, and time duration for better results. The applied algorithm gained 97% accuracy on the real rice image dataset during the training phase
Efficient and Accurate Image Classification via Spatial Pyramid Matching and SURF Sparse Coding: Efficient and Accurate Image Classification via Spatial Pyramid Matching and SURF Sparse Coding
No full textSparsely coded signal space representations do well in feature quantization. Instead of using standard vector quantization, the suggested method uses selective sparse coding to assemble the most important features of the appearance descriptors of nearby image patches. Inadequate coding also enables adjacent max pooling on some spatial scales, which, unlike the setup of average pooling in a histogram, links interpretation with scale invariance. The acquired visual illustration is the key contribution of this research; it performs outperform with linear-SVM, improves the model training's, which in turn speeds up testing with improves accuracy. The efficacy of the method we have employed has been substantiated through a series of experiments conducted on diverse datasets. Since top-performing image classification systems heavily rely on nonlinear SPM in mean of vector quantization, the trustworthy recommended linear SPM greatly increases the use of larger sets of training data. The method given herein deduces that the sparse coding of SURF feature’s function hampered a more comprehensive local appearance descriptor for general-purpose image processing. Experiments and comparisons are conducted on standard datasets such as Caltech-101, FTVL, and Corel-1000, using state-of-the-art techniques and descriptors. When compared over several other image categories and descriptors, the method provided here comes out on top
A Machine Learning Based Approach for the Detection of DDoS Attacks on Internet of Things Using CICDDoS2019 Dataset - PortMap
Full text linkIn today's technological era, the Internet has become ubiquitous, playing a vital role in our daily lives. With the exponential growth of IoT innovation, millions of interconnected IoT-enabled devices rely on cloud services to communicate over the Internet. However, this rapid development also exposes these devices to various threats, with DDoS (Distributed Denial of Service) and DoS (Denial of Service) attacks being particularly potent and destructive. DDoS attacks present a unique challenge as they are extremely difficult to detect using conventional intrusion detection frameworks and traditional methodologies. Fortunately, advancements in machine learning have provided a promising solution by enabling accurate differentiation between DDoS attacks and other forms of data. This study proposes a DDoS detection model based on machine learning algorithms. To conduct this study, we utilized the most recent and freely available online dataset called CICDDoS2019. Various machine learning-based techniques were explored, aiming to identify the characteristics associated with accurate classification. Among the algorithms tested, AdaBoost and XGBoost demonstrated exceptional performance. As part of future work, a hybrid approach will be incorporated into this model, further improving its capabilities. It is worth noting that this model will be continuously updated with new data on DDoS attacks, ensuring its relevance and effectiveness in combating emerging threats. By leveraging machine learning techniques, this approach enhances the detection of DDoS attacks on Internet of Things networks, safeguarding the integrity and security of connected devices and the overall IoT ecosystem
Strategic Customer Segmentation: Harnessing Machine Learning For Retaining Satisfied Customers
Full text linkThis research paper explores the burgeoning field of machine learning and its application in strategic customer segmentation within the aviation industry. Leveraging the Airline Passenger dataset, this study assesses the potential of various machine learning classifiers to enhance customer retention by effectively segmenting satisfied customers. Our methodology involves a comparative analysis of five machine learning classifiers: Random Forest, K-Nearest Neighbors (KNN), Decision Tree, Naive Bayes, and Artificial Neural Network (ANN). Each classifier is rigorously tested and evaluated based on key performance metrics including accuracy, precision, recall, and F1-score.
The results indicate a diverse range of classifier effectiveness. Notably, the Random Forest classifier outperforms others with outstanding metrics: accuracy, precision, recall, and F1-scoreof 0.96. Decision Tree follows closely, also achieving high performance with a score of 0.95 across all metrics. Naive Bayes and ANN demonstrate respectable performance, with accuracyscores of 0.86 and 0.90 respectively. In contrast, KNN presents lower but consistent performance, with all metrics at 0.75. These quantitative findings highlight the nuanced performancedifferences among classifiers, emphasizing the critical role of algorithm selection in achieving precise customer segmentation.
This study provides significant insights into the application of machine learning for strategic customer retention in the aviation sector, presenting practical implications for airlines aiming to optimize their segmentation strategies and retain satisfied customers. By showcasing the varying performances of different classifiers, this research contributes to the broader discourse on integrating machine learning into customer-centric strategies, ultimately aiding airlines to engage and retain their customer base more effectively