International Journal on Recent and Innovation Trends in Computing and Communication
Not a member yet
8613 research outputs found
Sort by
Predictive Analysis of Covid-19 Disease Severity in X-ray images: using Deep Learning Techniques
Get PDFHealthcare systems are evolving in order to deal with the issues of death in human. The most current worldwide pandemic, COVID-19, which first appeared in 2019, has spread throughout the world. Covid sickness is currently one of the leading causes of death in humans. The signs of COVID-19 include fever, coughing, exhaustion, body pains, and shortness of breath. These symptoms can range in severity from moderate to severe. Also possible for some people are sore throats, congestion, runny noses, and loss of taste or smell. The COVID-19 pandemic has prompted researchers to create imaging-based medical treatments, allowing medical staff to detect COVID-19-infected patients more quickly and begin necessary treatments on schedule. The new coronavirus (COVID-19) illness is extremely contagious, thus there are often too many patients waiting in line for chest X-rays. This burdens the radiologists and physicians and has a detrimental impact on the patient's treatment and pandemic management. Due to this highly contagious condition, there aren't as many clinical amenities available, such as hospitals with critical care units and ventilatory machines, it is now crucial to categorise the patients according to their severity levels. Using deep learning techniques, we categorized the individual based on the severity levels of moderate, severe, and extreme if they tested positive for COVID-19. The COVID-19 patient severity divided into three groups: moderate, serious, and extreme, using Convolution Neural Network (CNN) three architecture: VGG19, ResNet-50 and DenseNet201 model that was constructed with an average accuracy of VGG19-89.63%, ResNet-50 with 92.62% and DenseNet201 with 96.4% with the input of chest X-ray pictures
A Profound Multitask System for Gender Identification face recognition, Confront Discovery, Point of interest Localization, and Head Position Estimation Hyperface
Get PDFMachine learning is a technology that has risen in its usage and popularity in the last few years. A huge number of people from around the world are learning this technology and putting the knowledge to various use. Machine learning algorithms are capable of learning from the provided data with high accuracy. Even though a significant amount of research has been conducted on face recognition, the integrated model of face recognition, landmark localization, head posture estimation, and gender identification that is capable of high accuracy and speed has not yet been investigated. As a result, we have developed a face recognition system that can make predictions about photos that are comparable to those made by humans. The principal component analysis PCA and the SVM were used here to accomplish facial recognition. In feature extraction, to reduce the dimensionality of large datasets, principal component analysis is performed. After the data have been preprocessed, they are entered into the SVM classifier to be used for image classification. The study of this is done via visualization, and it is used to measure the effectiveness of the model. This face recognition algorithm has an accuracy of at least 80% when it comes to classifying people's portraits. The findings of the experiments show that the suggested technique can successfully identify faces since it employs a feature-based algorithm that combines PCA classification and SVM detection
Harnessing Convolutional Neural Networks for Histopathological Breast Cancer Classification.
Get PDFRecent advancements in Convolutional Neural Networks (CNNs) have significantly supported the field of breast cancer discovery using medical imaging. An improvised DenseNet architecture for the classification of histo-pathological breast cancer images is explored in this work. Leveraging the effectiveness of DenseNet in capturing intricate patterns through dense connectivity, our improvised architecture aims achieve high accuracy and efficiency of classification. The model integrates novel features such as optimized bottleneck layers and attention mechanisms, contributing to improved feature extraction and classification capabilities. The improvised DenseNet produced a accuracy of 93.39% on the breakhis dataset. A summary of key findings and future research directions, emphasizing the need of custom CNN models in breast cancer detection is provided
Ensemble Approach for DDoS Attack Detection in Cloud Computing Using Random Forest and GWO
Get PDFWhen multiple technologies are added to a traditional network, it becomes increasingly difficult to meet newly imposed requirements, such as those regarding security. Since the widespread adoption of telecommunication technologies for the past decade, there have been an enhancement in the number of security threats that are more appealing. However, many new security concerns have arisen as a consequence of the introduction of the novel technology. One of the most significant of these is the potential for distributed denial of service attacks. Therefore, a DDoS detection method based on Random Forest Classifier and Grey Wolf Optimization algorithms in this work was developed to mitigate the DDoS threat. The results of the evaluation show that the Random Forest Classifier can achieve substantial performance improvements with respect to 99.96% accuracy. Comparison is also made to several state-of-the-art techniques for detecting of DDoS attacks for the real dataset
Quantum Blockchain: Unraveling the Potential of Quantum Cryptography for Distributed Ledgers
Get PDFThe examination investigates the joining of quantum-safe cryptographic calculations into blockchain innovation, zeroing in on grid-based cryptography and hash-based marks. Because of the inescapable danger presented by quantum processing, this study proposes a quantum-safe blockchain system intended to upgrade the security and flexibility of circulated records. The cross-section-based cryptography calculation uses the computational intricacy of grid issues, offering protection from quantum goes like Shor's calculation. Simultaneously, hash-based marks give lightweight and quantum-safe choices for advanced marks, supporting the general validity of blockchain exchanges. The examination includes a multi-staged approach, incorporating a complete writing survey, hypothetical system improvement, algorithmic execution, and exhaustive investigation of versatility, execution, and information security. Reproduction results will illuminate ensuing equipment executions, approving the down-to-earth attainability of the proposed quantum-safe blockchain. Besides, the review digs into moral and administrative contemplations, adding to the foundation of capable rules for quantum-safe blockchain innovation. Insights into the performance of lattice-based cryptography and hash-based signatures, as well as the provision of a blueprint for future research in quantum-resistant distributed ledger systems, are among the anticipated contributions. The powerful idea of quantum advancements and blockchain requires continuous investigation, and the exploration makes way for future examinations concerning quantum-safe agreement components, upgraded Quantum Key Dispersion, and interdisciplinary coordinated efforts
Exploring Privacy-Preserving Methods via Perturbation Data Mining Employing Diverse Noise Strategies
Get PDFKnowledge discovery from data, commonly referred to as data mining. it involves the extraction of significant information, which may be previously unknown, concealed, or relevant, from extensive data sets or databases through the utilization of statistical methodologies. With the introduction of enhanced hardware technologies, there has been a proliferation in the storage and recording of personal data pertaining to individuals. Sophisticated organizations employ data mining algorithms to uncover hidden patterns or insights within data. Data mining techniques find application in diverse fields such as marketing, medical diagnosis, forecasting system, and national security. However, in scenarios where data privacy is paramount, mining certain types of data without violating the privacy of data owners presents a formidable challenge, sparking growing concerns among privacy advocates. To address these concerns, it is imperative to advance data mining procedures that are complex to individual privacy considerations. Perturbation of data plays a pivotal role in Privacy-Preserving Data Mining (PPDM). Additive data safeguard data privacy. In contrast, multiplicative data perturbation involves a series of transformations, including rotation, translation, and the addition of noise components to the perturbed data copy
Enhancing the Security and Privacy of eHealth Records through Blockchain-based Management: A Comprehensive Framework
Get PDFProgress in information technology is transforming the healthcare sector with the goal of enhancing medical services, diagnostics, and continuous monitoring through wearable devices, among other benefits, while also lowering expenses. This digital transformation enhances the convenience of computing, storing, and retrieving medical records, ultimately leading to improved treatment experiences for patients. Electronic health record systems have come under fire for centralized control, faults, and attack points with transferring data custodians. These systems are frequently utilized for the interchange of health information among healthcare stakeholders. The main objective is to overcome information asymmetry and data breaches commonly encountered in the Electronic Health Record (EHR) system. This study introduces a decentralized and trustless architecture aimed at securely storing patients' medical records and granting access to authorized individuals, including healthcare providers and patients themselves. The research primarily focuses on bolstering the security and privacy of healthcare data management systems using blockchain technology. To address the issue of blockchain scalability, an off-chain scaling approach is proposed, utilizing an underlying medium to store large volumes of data. This is achieved through the integration of Elliptic Curve Cryptography (ECC) and the Interplanetary File System (IPFS). The proposed system provides a secure and efficient method for storing and sharing sensitive healthcare data while ensuring confidentiality and data integrity
High Isolation Wideband MIMO Antenna without Decoupling Technique for IoT Applications
Get PDFThis paper presents a method a high isolation wideband MIMO antenna without using any decoupling technique. This is achieved by transforming a strip line to a 1 ×2 MIMO antenna using Hilbert transform and Defected Ground structure is used in the antenna to resonate the antenna in for IOT and 5G sub-6 GHz bands. The proposed antenna has a size of 50 × 49.8 × 1.6 mm 3 and operates in the frequency band of 4.6 GHz to 5.94 GHz. The antenna simulated in ANSYS HFSS showed that its parameters Envelope correlation coefficient (ECC), Total Active Reflection Coefficient (TARC), Diversity gain (DG) and Channel Capacity Loss (CCL) are less than 0.04, -25 dB, 9.98 to 10 dB and less than 0.4 bits/s/Hz respectively. The radiation pattern of the antenna in both E-plane and H-plane has been simulated which the uniform distribution of power in the space
Significance of Data Structures and Data Retrieval Techniques on Sequence Rule Mining Efficacy
Get PDFSequence mining intends to discover rules from diverse datasets by implementing Rule Mining Algorithms with efficient data structures and data retrieval techniques. Traditional algorithms struggle in handling variable support measures which may involve repeated reconstruction of the underlying data structures with changing thresholds. To address these issues the premiere Sequence Mining Algorithm, AprioriAll is implemented against an Educational and a Financial Dataset, using the HASH and the TRIE data structures with scan reduction techniques. Primary idea is to study the impact of data structures and retrieval techniques on the rule mining process in handling diverse datasets. Performance Evaluation Matrices- Support, Confidence and Lifts are considered for testing the efficacies of the algorithm in terms of memory requirements and execution time complexities. Results unveil the excellence of Hashing in tree construction time and memory overhead for fixed sets of pre-defined support thresholds. Whereas, TRIE may avoid reconstruction and is capable of handling dynamic support thresholds, leading to shorter rule discovery time but higher memory consumption. This study highlights the effectiveness of Hash and TRIE data structures considering the dataset characteristics during rule mining. It underscores the importance of appropriate data structures based on dataset features, scanning techniques, and user-defined parameters
Hydrological Modeling of Large River Basin Using Soil Moisture Accounting Model and Monte Carlo Simulation
Get PDFThis description outlines a Geographic Information System (GIS)-based rainfall-runoff model that simulates the flow of water in a river basin. The model operates on a daily time step and consists of four non-linear storage components: interception, soil moisture, channel, and groundwater. It employs (SCS) Unit Hydrograph model to determine unit hydrograph ordinates. The model replicates the movement and storage of water in various parts of the basin, including vegetation, the soil surface, the soil profile, and groundwater layers. To address uncertainty, a Monte Carlo simulation feature is integrated into the model. This feature generates required number of sample sets with random parameter values. The model is run for all these realizations during a calibration period, and performance metrics like NSE are calculated for each calibration yearTo assess prediction uncertainty, model parameter weights are computed by normalizing the corresponding likelihood values. These weights sum up to one and represent the probabilistic distribution of predicted variables, illustrating the impact of structural and parameter errors on model predictions. A sensitivity analysis reveals that the Muskingum constants K and X have the greatest influence on model performance, while parameters ?GW, ?SW, ?fc, and ?pc have a minimal effect on the model's performance