Authors: Asst. Prof Anil Choubey

Abstract: This paper presents the estimation of harmonics in a voltage source converter based HVDC (VSC-HVDC) system for designing AC side filters. The extended VSC and PSO is well known for estimating amplitude, phase, frequency, and harmonic content of a signal corrupted with noise. However, the algorithm suffers from instability due to linearization and costly calculation of Jacobian matrices, and its performance deteriorates when the signal model is highly nonlinear. This paper, therefore, proposes an unscented to overcome these difficulties of linearization and derivative calculations for robust tracking of harmonics in VSC-HVDC system. The model and measurement error covariance matrices Q and R along with the VSC parameters are selected using a modified particle swarm optimization (PSO) algorithm. To circumvent the problem of premature convergence and local minima, a dynamically varying inertia weight based on the variance of the population fitness is used. This results in a better local and global searching ability of the particles, which improves the convergence of the velocity and better parameters. Various simulation results for harmonic signals corrupted with noise obtained from VSC-HVDC system reveal significant improvement in noise rejection and speed of convergence and accuracy.

Published by: vikaspatanker

Authors: Suman Lata, Dr. Upendra Kumar Srivastava

Abstract: Generative models such as GANs and diffusion systems enable the creation of highly realistic fake videos, eroding confidence in online content across social, political, and legal domains. Synthesizing insights from more than 85 scholarly articles published between 2018 and 2025, this work categorizes detection methods into spatial CNNs that identify frame-level flaws, temporal RNNs/LSTMs for motion inconsistencies, RPG-based physiological cues, and fused audio-video approaches. Evaluations on datasets like Celebs and Wild Deepfake yield accuracies above 95% in controlled settings, but cross-dataset generalization and defenses against advanced forgeries falter. Hybrid architectures with transformers emerge as leaders, revealing critical gaps in real-time efficiency and edge-device applicability to steer forthcoming innovations.

DOI: https://doi.org/10.5281/zenodo.18253039

 

Published by: vikaspatanker

Authors: Vishrut Nath Jha, Joanne Anto, Athira KK

Abstract: Recent advancements in vector databases and embedding-based retrieval have transformed how large unstruc- tured datasets are indexed and searched. However, the perfor- mance of emerging storage and retrieval systems varies widely depending on their architectural design and optimization goals. This study presents a comparative evaluation of three distinct approaches MemVid, Qdrant, and Amazon S3 Vectors using a dataset of 10,417 medical-text chunks derived from research documents. Each system was assessed in terms of indexing efficiency, retrieval accuracy, latency, and resource utilization. Experimental results demonstrate that MemVid, which stores em- beddings in a video-encoded FAISS-based format, achieved lower query latency and higher retrieval precision for this corpus, while Qdrant exhibited superior scalability and flexibility in handling dynamic updates and metadata filtering. Amazon S3 Vectors, though currently in preview, offered cloud-native durability and seamless AWS integration with moderate performance overhead. The analysis reveals that no single vector system universally outperforms others; rather, each excels under specific workload conditions. These findings provide practical guidance for selecting appropriate vector storage backend based on corpus size, update frequency, and deployment environment.

DOI: http://doi.org/10.5281/zenodo.18243168

Published by: vikaspatanker

Authors: Lekshmi O Nair

Abstract: The equatorial and low-latitude ionosphere over the Indian region has undergone significant changes during the past 25 years (2000-2025), spanning three complete solar cycles. This comprehensive review synthesizes observational evidence from ground-based ionosondes, GPS Total Electron Content (TEC) measurements, satellite missions, and space weather monitoring networks across the Indian longitude sector (60°E-100°E). Key findings reveal systematic variations in the Equatorial Ionization Anomaly (EIA) morphology, with the northern crest showing enhanced TEC values during solar maximum periods and distinctive seasonal asymmetries. Equatorial Spread F (ESF) and plasma bubble occurrence demonstrate strong correlations with solar flux variations, geomagnetic activity, and atmospheric tidal forcing. The review highlights technological advances including the Indian Network for Space Weather Impact Monitoring (InSWIM), improved ionospheric models, and enhanced prediction capabilities. Significant space weather events, including the 2003 Halloween storms, 2015 St. Patrick's Day event, and recent solar cycle 25 disturbances, have provided insights into ionosphere-magnetosphere coupling processes. Climate change effects on the upper atmosphere are emerging as new research frontiers, with evidence for long-term trends in ionospheric parameters. Future challenges include understanding mesosphere-lower thermosphere-ionosphere coupling, developing regional ionospheric models for navigation applications, and preparing for increasing space weather threats to technological infrastructure.

DOI: http://doi.org/10.5281/zenodo.18242035

Published by: vikaspatanker

Authors: Dev Dharrshan S, Divya Dharshini S, Anusree D, HemaPriya Vs, Avaneesh R, Deshika Sri KM, Dhavanithi M

Abstract: Smart Tourist Safety refers to the integration of digital technologies such as the Internet of Things (IoT), mobile applications, artificial intelligence, and real-time data analytics to enhance the safety and security of tourists. With the rapid growth of global tourism, ensuring tourist safety has become a critical concern for destinations and governments. Smart safety systems enable real-time monitoring, emergency response, location tracking, and risk prediction, helping tourists navigate unfamiliar environments securely. This study explores the concept of Smart Tourist Safety, examines key technologies involved, and discusses their role in improving emergency management, crime prevention, and overall tourist confidence. The findings highlight that smart safety solutions not only reduce risks but also enhance destination attractiveness and sustainability.

Published by: vikaspatanker

Authors: Dr. Kumari Priyanka

Abstract: Role of a secondary alcohol (Propan-2-ol) in affecting the kinetics and the thermodynamic extensive properties of the acid catalysed solvolysis of iso-butyl formate has been highlighted by studying the kinetics of the reaction in different aquo-propan-2-ol reaction media having 20 to 80% (v/v) of propan-2-ol and at five different temperatures rising from 20 to 40°C at interval of 5°C It was found that with increase in Jemperature of the reaction from 20 to 40°C from 0.172 to 1.344 molecules of water are associated with the activated complex and from this, it is inferred that mechanistic path followed by the reaction in presence of propan-2-ol is changed from bimolecular to unimolecular. The depletion and enhancement observed respectively in iso-composition and iso-dielectric activation energies reveal that the transition state is solvated and initial state is desolvated with addition of propan-2-ol in reaction media. Almost unity value of the slope of the plots of log k values against log [H+] values shows that the reaction follows AAC2 mechanism. From the values of iso-kinetic temperature, which comes to be 287, it may be concluded that in water-propan-2-ol reaction media, the reaction follows Barclay-Butler rule and there is weak but acceptable interaction between solvent and solute in aquo-propan-2-ol reaction media

DOI: https://doi.org/10.5281/zenodo.18373220

 

Published by: vikaspatanker

Authors: Gayatri Solanki, Abhay Mundra

Abstract: Gender identification has gained significant attention in recent years because it supports many real-world applications related to demographic analysis and human-centered systems. Gender classification refers to the automated process of predicting a person’s gender (typically male or female) based on visual appearance, most commonly from facial images. However, extracting reliable and discriminative facial features remains challenging due to variations in lighting, pose, facial expressions, occlusion, and image quality. With advances in machine learning and deep learning, automatic gender classification systems have become increasingly accurate and widely adopted across multiple domains. These systems can be useful in security and access-control environments, as well as in demographic analytics and personalized services. In certain contexts, gender identification may also be applied to manage access in gender-specific spaces and services, such as women-only transportation sections or gender-segregated facilities. This review summarizes key traditional machine learning approaches (e.g., SVM-based methods) and modern deep learning techniques (e.g., convolutional neural networks), and discusses commonly used benchmark datasets and evaluation practices for gender classification.

Published by: vikaspatanker

Authors: Devendra Gupta, Abhay Mundra

Abstract: Brain tumor segmentation is a critical task in medical imaging, supporting accurate diagnosis, treatment planning, and continuous monitoring of tumor progression. Over the years, a variety of segmentation strategies have been proposed, each offering distinct advantages and limitations. Early traditional approaches—including thresholding, edge-based detection, and region-growing methods—are computationally efficient and simple to implement, but they often perform poorly in the presence of noise, intensity inhomogeneity, and complex or ambiguous tumor boundaries. Statistical and model-driven techniques, such as clustering methods and deformable models, improve adaptability to anatomical variability but typically require careful parameter selection and may involve higher computational cost. In recent years, machine learning and deep learning methods have transformed brain tumor segmentation, particularly through the use of Convolutional Neural Networks (CNNs) and U-Net-based architectures, which have demonstrated strong accuracy and robustness across large and diverse MRI datasets. More recently, hybrid methods that combine classical image processing with deep learning have gained attention for improving efficiency, interpretability, and generalization. This review summarizes the evolution of brain tumor segmentation methods, compares major categories of approaches, and discusses current challenges and promising future research directions.

Published by: vikaspatanker

Authors: Devendra Gupta, Abhay Mundra

Abstract: Brain tumor detection from magnetic resonance imaging (MRI) is essential for early diagnosis, treatment planning, and improved patient outcomes. This study conducts a comparative evaluation of five deep learning approaches—CNN, ResNet50, U-Net, YOLOv7, and InceptionResNetV2—for automated brain tumor classification. Prior to model training, the dataset was prepared through systematic preprocessing, including data cleaning, normalization, and augmentation to improve robustness and reduce overfitting. Model performance was assessed using standard classification metrics: accuracy, precision, recall, and F1-score. Experimental results indicate that all evaluated architectures achieved strong predictive performance; however, InceptionResNetV2 consistently outperformed the other models, achieving near-perfect scores across all evaluation measures. This strong performance suggests improved reliability in reducing both false-positive and false-negative predictions, making InceptionResNetV2 a promising candidate for clinical decision-support applications. Overall, the findings highlight the importance of advanced deep learning architectures in delivering accurate and dependable MRI-based brain tumor detection.

Published by: vikaspatanker

Authors: Gayatri Solanki, Abhay Mundra

Abstract: Gender identification from facial images is an important problem in computer vision and has attracted growing interest due to its applications in surveillance, security, and human-centered systems. Although humans can infer gender naturally, developing automated systems that perform reliably across real-world conditions remains challenging. This work presents a gender classification framework that leverages face recognition feature vectors for prediction. First, face images are detected, aligned, and preprocessed to obtain a standardized facial representation. Next, a face recognition network extracts compact feature embeddings that encode discriminative facial characteristics. Finally, machine learning and deep learning classifiers are applied to these embeddings to determine gender. The proposed system integrates advanced components including VGG-Face, Deep Belief Networks, and shifted filter responses to improve robustness. Multiple deep learning architectures were investigated—CNN, VGG16, ResNet50, InceptionV3, and EfficientNet—with ResNet152 showing the strongest overall performance. Experimental findings indicate that ResNet152 achieves approximately 9% improvement over leading alternatives and demonstrates enhanced resilience to anomalies and variations compared with earlier approaches.

Published by: vikaspatanker