2025

Permanent URI for this collectionhttps://dl.bhu.ac.in/bhuir/handle/123456789/62057

Browse

Now showing 1 - 20 of 50

A comparative overview of Riemannian and Finsler geometry
(American Mathematical Society, 2025) Bankteshwar Tiwari
The aim of this article is to present a comparative overview of Riemannian and Finsler geometry, starting from some historical developments to the various directions of current research. This includes the discussion on classification of Finsler spaces of constant flag curvature and constant Ricci curvature, cut locus, conjugate locus, Comparison theorems, Gauss-Bonnet-Chern Theorem and sphere theorem in Finsler geometry. The topological, differential and metric structures on Riemannian manifolds in the presence of convex functions have been active fields of research in the second half of the last century. We discuss some of these results on Riemannian manifolds with convex functions and their recently extended analogues on Finsler manifolds. © 2025 American Mathematical Society.
A Comprehensive Review of Credit Card Fraud Detection Algorithms: Merits, Challenges and Future Directions
(Springer Science and Business Media Deutschland GmbH, 2025) Nitish Kumar Mishra; Rakhi Garg
Fraud is a serious crime that affects both financial institutions and individuals badly. The growth of the digital world and ease of access to the internet has led to the increase in fraudulent activities tremendously. With the evolution of technology, the methods of frauds has also evolved. In this paper we focuses on credit card fraud detection algorithms. Credit card frauds are increasing because of its wide usage and huge amount of credit limit. Credit cards are ubiquitous and integral part of online financial transactions. In this paper we mainly focuses on various credit card fraud detection algorithms stating their merits and demerits and also highlights the various issues and challenges in the various algorithms developed so far. Though Data Mining and Machine Learning techniques are used for credit card fraud detection but still there is a need of efficient technique for real time fraud detection. Ensemble learning, Deep Learning techniques works better in detecting real time fraud. Lack of availability of datasets, class imbalance in datasets, feature selection are major challenges in detecting credit card fraud. The study is useful to the researchers and scientists working in the area of credit card fraud detection algorithms. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
A Dual Adaptation Approach for EEG-Based Biometric Authentication Using the Ensemble of Riemannian Geometry and NSGA-II
(Springer Science and Business Media Deutschland GmbH, 2025) Aashish Khilnani; Jyoti Singh Singh Kirar; Ganga Ram Gautam
Recently, it has been discovered that EEG signals have enormous potential to be used as biometric authentication. Although, its practical implementation is limited due to the intricate and dynamic nature of EEG signals. To overcome these challenges, we need to simplify the analysis and preserve the spatial attributes of the EEG signals. In this work, a methodology using an ensemble of Riemannian geometry and a genetic algorithm for EEG-based biometric authentication is devised. Here, the symmetric positive definite covariance matrices of the EEG signals are calculated and classified using the Minimum distance to the Riemannian Mean (MDRM) and the Tangent space LDA (TSLDA) classifier. Furthermore, NSGA-II is used to optimize the number of channels and to reduce the computational complexity. This study achieved an accuracy of 99.9% on average with all the datasets used. Multiple publicly available datasets are used to compare the proposed approach with existing methods. Results obtained show the efficacy of the proposed method. Friedman’s statistical test also supports the statistical significance difference between the proposed and existing methods. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
A Hybrid Approach for Preserving Source Location Privacy for Wireless Sensor Networks
(Springer Science and Business Media Deutschland GmbH, 2025) Nisha Singh; S. Suresh
The advancing development in wireless sensor networks (WSNs) has led to its immense applications in location-based services. However, along with various advantages it suffers from privacy concerns. The protection of location information of the sensor node that reports the occurrence of an event in its sensing range is very important. In this paper, we propose a hybrid approach of randomized and angular routing that is guaranteed at preserving the location of the source. The proposed scheme makes use of fake packets to create a time discrepancy in backtracking the packet flow that increases difficulty for the adversaries. Even though the use of fake packets to create traffic leads to a slight increase in energy consumption, it has also provided a higher level of security to source node. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
A Novel Hybrid Islanding Detection Technique for PV-Battery DC Microgrid
(Institute of Electrical and Electronics Engineers Inc., 2025) Kunal Singh; Avirup Maulik; Mitresh Kumar Verma; Soumya R. Mohanty
A hybrid islanding detection technique for a DC microgrid is presented in this paper, in which a novel passive index (cumulative product of superimposed voltage) is proposed for disturbance identification in the first stage (passive). The second (active) stage discriminates between islanding and nonislanding disturbances and is activated if the proposed passive index exceeds a threshold. The computation of the proposed passive index is simple, requiring only voltage measurements, making the proposed approach fast and easily implementable. Simulation studies verify that the proposed technique can successfully discriminate between islanding and non-islanding events, quickly detect islanding under zero power mismatch scenarios, and improve power quality. © 2025 IEEE.
Addressing Graphviz File Generation Issue in CPN Tools: A Java-Based Solution
(Springer Science and Business Media Deutschland GmbH, 2025) Vikas Kumar Patel; Anshul Verma; Pradeepika Verma
This paper presents a novel approach to resolving a specific bug encountered in Colored Petri Nets (CPN) Tools during the generation of Graphviz files. CPN Tools, a powerful instrument for constructing and analyzing colored petri nets, has been identified to have limitations in graph visualization, particularly when dealing with token value strings or record/product with strings.The CPN Tools generate the Graphviz (dot) file, however it does not take into account color sets of strings, records, and products. If the color set is one of these, after generating the Graphviz file from CPN, Graphviz displays a syntax error. This issue results in a syntax mismatch with Graphviz, hindering the seamless visualization of graphs. To address this problem, the authors have developed a Java-based Graphical User Interface (GUI) program that takes the CPN Tools generated dot file as input, refactors it, and corrects the syntax, thereby enabling error-free graph visualization. The user merely needs to provide the CPN Tools created DOT file to the proposed system. It will make internal corrections and generate the graph. In addition, the user can download the revised version of the DOT file from the system. This solution not only resolves the identified bug but also enhances the overall user experience by simplifying the process of graph generation. The paper provides a detailed description of the bug, the development process of the Java GUI program, enhancing the page hierarchy graph of hierarchical colored petri nets, and its successful application in resolving the bug and generating correct graph. It also discusses the potential impact of this solution on the broader usability and efficiency of CPN Tools. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
Advanced Brain Tumor Classification from MRI Images with Vision and Swin Transformer Models
(Institute of Electrical and Electronics Engineers Inc., 2025) Anjali Jain; Ankita Mishra; Bethany Gosala; Manjari Gupta; Alwin Poulose
Brain tumors, characterized by abnormal cell growth within the brain, represent one of the most life-threatening conditions due to their potential to severely impair brain function and cause neurological complications, including death. Accurate classification of brain tumors is crucial in medical diagnosis, as misdiagnosis can lead to ineffective treatment and reduced patient survival. Magnetic Resonance Imaging (MRI) is a widely utilized non-invasive technique for acquiring high-contrast grayscale brain images commonly employed in tumor diagnosis. This study uses MRI images to evaluate the effectiveness of two pre-trained transformer-based models for brain tumour classification, specifically the Vision Transformer and Swin Transformer. The models were trained and tested on a publicly available Kaggle dataset containing 3,000 brain MRI scans. Our methodology employed the fine-tuned Vision transformer and Swin Transformer on the MRI dataset for classification. The models' performance was assessed using accuracy, precision, recall, and F1-score metrics. The Vision Transformer achieved an accuracy of 95.30%, while the Swin Transformer outperformed it with an accuracy of 98.71%. These results highlight the efficacy of transformer models in accurately identifying common brain tumors, demonstrating their potential to enhance computer-aided diagnosis and support healthcare professionals in making swift and accurate diagnostic decisions. © 2025 IEEE.
Analysis of Source and Source Region of Coarse Mode Aerosol (PM10) in Varanasi, India
(Springer Science and Business Media Deutschland GmbH, 2025) Preeti Tiwari; Bharat Ji Mehrotra; Manoj Kumar Srivastava; Manoj C.Anil Kumar; Narayanasamy Vijayan; Sudhir Kumar Sharma
In this study, we conducted source apportionment and source region analysis of elemental concentrations of coarse mode aerosol (PM10) using principal component analysis (PCA), trajectory analysis, and conditional bivariate probability function (CBPF) in Varanasi, situated over central Indo-Gangetic Plain (IGP) of India. In this case, 34 elements (Mg, Na, Al, Si, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Ga, Zr, Mo, Ag, Pm, Y, Rb, Pd, Fr, U, Pb, B, P, S, Cl, Br, Nb, Sr, F, As, and Ba) were extracted from PM10 samples using WD-XRF technique. PCA identified six sources (crustal/soil/road dust, biomass burning, industrial emissions, agricultural activities, combustion, and vehicular emissions) of elemental concentrations of PM10 over Varanasi. Trajectory analysis indicated contribution to PM10 concentration transported from Uttar Pradesh, Punjab, Haryana, Delhi, partially from Rajasthan, Gujarat, West Bengal, and some parts of Madhya Pradesh, with additional transboundary emissions from Pakistan, Afghanistan, and Nepal. CBPF analysis predicted the local PM10 emissions from North-East, South, West, and North-West directions, with major contributors including Ganga ghats, religious places, market areas, and traffic activities. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Applicability of CoRTN Model in Indian Road Traffic Condition
(Springer Science and Business Media Deutschland GmbH, 2025) Ashish Kumar Chouksey; Brind Kumar; Manoranjan Parida; Amar Deep Pandey
The primary step in reducing traffic noise is to build a model for traffic noise prediction. The suitability and applicability of calculation of road traffic noise (CoRTN) model were checked in Indian city road traffic condition and national highways. Both roads were different from each other in many ways. Previously CoRTN model was successfully applied in homogeneous traffic condition, but its applicability in heterogeneous traffic flow requires much attention. Four locations were selected, and two locations were on homogenous traffic flow condition (national highways) and another two location on heterogeneous traffic flow (city roads). The range of noise level (L10) in national highways were 76–80 dBA and in city roads were 80–84 dBA. The result of the study showed that CoRTN model predicted well for national highways with R2 value 0.817. While for city roads, model fits the data really poorly. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Bridging FAIR and CARE in ETD Metadata: An LLM-based Cross-Repository Evaluation Framework
(Dublin Core metadata initiative, 2025) Somesh Rai; Dr Rajani Mishra
Electronic Theses and dissertations that are submitted electronically are known as electronic theses and dissertations (ETDs), and they are distributed through a variety of institutional and aggregated repositories. Even though the majority of ETD platforms adhere to the FAIR principles, which specify that data must be Findable, Accessible, Interoperable, and Reusable, these platforms frequently fail to take into account the ethical and community-centered aspects that are encapsulated in the CARE principles, which are as follows: Collective Benefit, Authority to Control, Responsibility, and Ethics. The purpose of this research is to present a novel cross-repository evaluation framework that utilises an LLM-assisted technique to bridge the gap between the FAIR and CARE principles. A rubric-based review was combined with the reasoning skills of three big language models-ChatGPT, Grok, and DeepSeek R1-in order to conduct an evaluation of nine of the important open-access electronic text databases (ETD) repositories. A standardised rubric served as a guide for each model, and it was prompted to conduct an analysis of the quality of the metadata as well as ethical constraints. Despite the fact that the data demonstrate that FAIR compliance is resilient across repositories, they also highlight systemic weaknesses in CARE alignment, particularly with regard to cultural context, ethical reuse, and authorial control. Moreover, the comparative analysis among three agents suggests that it should be used for evaluating FAIR compliance. CARE compliance evaluation may need more sophisticated ‘Human in the Loop’ setup. This framework offers a scalable and transparent approach to analysing metadata governance. Additionally, it gives schema-agnostic recommendations for encouraging inclusivity and ethical stewardship in digital academic infrastructure. These characteristics are achieved through the triangulation of assessments given by artificial intelligence. © 2025 Copyright for this paper by its authors.
Data to Decisions: A Computational Framework to Identify Skill Requirements from Advertorial Data
(Springer Science and Business Media Deutschland GmbH, 2025) Aakash Singh; Anurag Kanaujia; Vivek Kumar Singh
Human capital, or skilled labor, stands out among the factors of production due to its ability to continually evolve and adapt to changing conditions and resources. This flexibility makes human capital the most essential factor for achieving sustainable growth in any industry/sector. As new technologies are developed and adopted, the new generations are required to acquire skills in newer technologies in order to be employable. At the same time professionals are required to upskill and reskill themselves to remain relevant in the industry. There is however no straightforward method to identify the skill needs of the industry at a given point of time. Therefore, this paper proposes a data to decision framework that can successfully identify the desired skill set in a given area by analysing the advertorial data collected from popular online job portals and supplied as input to the framework. The proposed framework uses techniques of statistical analysis, data mining and natural language processing for the purpose. The applicability of the framework is demonstrated on CS&IT job advertisement data from India. The analytical results not only provide useful insights about current state of skill needs in CS&IT industry but also provide decision support to prospective job applicants, training agencies, and institutions of higher education & professional training. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
Deep Learning Based EEG Based Mental Workload Detection with Discrete Wavelet Transform and Welch's Power Spectral Density
(Elsevier B.V., 2025) Shraddha Jain Sharma; Ratnalata Gupta
EEG data provide a convenient and portable method of measuring mental workload. This work analyzes EEG data collected during rest intervals and arithmetic problems using deep learning and machine learning approaches. EEG signals were broken down into sub-bands using the Discrete Wavelet Transform (DWT), and the power levels of these bands were estimated using Welch's approach. Based on these power levels, feature vectors were identified using both conventional machine learning techniques including Linear Discriminant Analysis (LDA), k-Nearest Neighbors (k-NN), and Support Vector Machines (SVM) and deep learning algorithms like Long Short-Term Memory (LSTM). The findings show that LSTM performed better than conventional classifiers, with higher performance metrics and accuracy. This demonstrates how deep learning may be used to separate mental workload states from EEG signals with accuracy. © 2025 The Authors. Published by Elsevier B.V.
Deep Learning for Cognitive Task and Seizure Classification with Hilbert–Huang Transform and Variational Mode Decomposition
(Springer Science and Business Media Deutschland GmbH, 2025) Shraddha Jain; Rajeev Srivastava
Electroencephalography (EEG) signals are often utilized to study cognitive processes and brain diseases. The non-stationary and non-linear nature of EEG signals makes their analysis difficult. A deep learning framework that suggested classifying seizures based on scalp EEG signals and automating cognitive tasks. We use a pre-processing module based on the Hilbert–Huang Transform (HHT) and Variational Mode Decomposition (VMD) to extract features from raw EEG data. We present an approach that combines deep learning with HHT and VMD for rapid and precise seizure detection. Our method detects seizures with an astounding 98.40% accuracy. Our suggested techniques have great potential for quantifying brain wave patterns and advancing neuroscience research, even outside of classification applications. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Design of High Gain Vivaldi Antenna with Substrate-Integrated-Waveguide Feed at L-Band
(Institute of Electrical and Electronics Engineers Inc., 2025) Soham Banerjee; Sougata Chatterjee; S. Sureshkumar; Gobinda Sen; Malay Ganguly; Somak Bhattacharyya
A high-gain Vivaldi antenna with a SubstrateIntegrated Waveguide (SIW) feed is proposed for L-band applications in this paper. The antenna operates over a frequency range of 1.05 GHz to 1.92 GHz, nearly covering the entire L-band spectrum. To minimize losses and enhance efficiency, an SIW feed has been integrated into the design. Additionally, an edge-slotting technique has been employed to improve radiation at lower frequencies. The antenna is fully metallic and has compact dimensions of 0.768 λ0 × 0.350 λ0 × 0.0035 λ0. It maintains a stable gain of 20-22 dBi across the entire operating band. The proposed design can be easily configured into antenna arrays, making it well-suited for radio astronomical applications and other high-gain, global navigational systems. © 2025 IEEE.
Effects of a Single-Session High-Intensity Interval Training on Blood and Liver Function Markers in Athletes
(Springer Science and Business Media Deutschland GmbH, 2025) Anshul Meena; Pradeep Singh Chahar; Jagdeep Singh
Regular physical activity is recommended for overall health, but many individuals struggle to meet these guidelines due to time constraints. High-intensity interval training (HIIT) offers a time-efficient alternative by incorporating short bursts of intense exercise with periods of rest. This study investigates the influence of a single-session HIIT on blood and liver function markers in adult male athletes, who may be at increased risk of adverse health outcomes due to the demands of rigorous training. Twenty male athletes (aged 22–26 years) were recruited from a university in India. Participants underwent a standardized HIIT protocol following a warm-up. Blood samples were collected at baseline and at intervals post-exercise, and analysed for various blood and liver function markers. Significant time-dependent changes were observed in haemoglobin (ES = 0.71, p < 0.05), white blood cell count (ES = 0.76, p < 0.05), red blood cell count (ES = 0.76, p < 0.05), platelet count (ES = 0.56, p < 0.05), and serum glutamic oxaloacetic transaminase (SGOT) levels (ES = 0.27, p < 0.05) as determined by repeated measures ANOVA. In contrast, no significant changes were observed in serum glutamic pyruvic transaminase (SGPT) levels (ES = 0.07, p > 0.05). Post-hoc analysis with Bonferroni correction revealed a significant increase in most parameters immediately following exercise, followed by a gradual return towards baseline levels within 20 min, except for platelet count. These findings provide valuable insights into the acute physiological responses of athletes’ blood and liver function markers to a single session of HIIT, emphasizing the importance of careful monitoring and effective recovery strategies following high-intensity training and further research to understand the long-term implications for athletic performance and overall health. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Enhanced Integral Sliding Mode Observer Integrating Super-Twisting Algorithm for Precision Control of PMSM in EV Drivetran
(Springer Science and Business Media Deutschland GmbH, 2025) Uppalapati Sudheer Kumar; Sukanta Halder; Naveen Yalla; Nilanjan Das; S. K. Bittu; Anand Kumar
The global demand for electric vehicles is increasing exponentially; on the contrary, the demand for the efficient drive motor control also paved a path for advanced motor control algorithms for smoother and dynamic control of modernday motors. The work presented in this chapter discusses about the controller specifically designed using the Super-Twisting algorithm incorporated with Integral Sliding Mode Observer (ST-ISMO), designed for Permanent Magnet Synchronous Motors (PMSMs). Addressing the limitations of traditional Proportional Integral (PI) controllers, the proposed ST-ISMO incorporates an enhanced sliding mode observer (SMO) with ST algorithm for mitigating chattering issues, which are most unavoidable effects in PMSM motor control. In addition, for addressing the issues associated with disturbances in the circuit of motor drive system both internal and external, an Extended State Observer (ESO)-based disturbance estimator is designed and incorporated into the proposed controller. The proposed controller is validated in MATLAB software, and the simulation results indicate that the ST-ISMO offers superior accuracy in control algorithm, speed ripples were minimized, and proposed controller settles the speed ripples within milliseconds time, when contrasted with traditional controllers. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Enriching Pre-Training Using Fuzzy Logic
(Institute of Electrical and Electronics Engineers Inc., 2025) Vansh Gupta; Vandana Bharti; Abhinav Kumar; Anshul Sharma; Sanjay Kumar Singh
Graph representation learning advances graph machine learning by encoding structural and relational information into feature vectors. This study introduces a fuzzy logic-based pre-processing layer that enhances node representations by adding semantic diversity and contextual understanding. The layer models uncertainty and captures abstract semantic characteristics in graph data, addressing the limitations of conventional methods that depend solely on structural attributes. By applying defuzzification, the layer refines embeddings, improving their robustness and effectiveness for a wide range of downstream tasks. To test its robustness, we introduce Gaussian noise ranging from 2% to 10% into the datasets, simulating real-world data imperfections. We evaluate the proposed layer on GraphMamba architecture, using DeepWalk and Node2Vec as baseline node feature generation algorithms. The results show consistent improvements in accuracy, F1 scores, precision, and recall across different noise levels. Our findings demonstrate the layer's ability to preserve high representational quality, speed up convergence, and handle noisy representations effectively. © 2025 IEEE.
Explainable Artificial Intelligence Paradigm in Cardiovascular Plaque Tissue Characterization and Classification for Risk Prediction
(Institute of Electrical and Electronics Engineers Inc., 2025) Saritha Lalitha Ravindran; Dipti S. Jadhav; Pankaj K. Jain
Atherosclerosis is the predominant cardiovascular disease caused by the blockage of blood vessels in various parts of the body which restricts the blood flow and can result in life-threatening conditions like stroke and heart attack. Several studies prove cardiovascular diseases are the major cause of fatality globally. Early detection of this precarious situation assumes importance in the proper administration of treatment methodologies. Artificial intelligence-based models can be effectively used for the segmentation, characterization, and classification of medical images to support clinicians in predicting the severity of blockages with improved accuracy and transparency. Conventional artificial neural network frameworks are perceived as 'black-boxes' as they are difficult to analyze and interpret by medical practitioners. Hence there is a growing interest in developing explainable 'white-box' techniques that are fully transparent and easily interpretable for medical and biological image processing to complement the existing approaches. This work covers a detailed literature review on how explainable artificial intelligence architectures enhance plaque segmentation, characterization, and classification. Also discusses the different modalities used for image acquisition, details of how model accuracy is calculated, limitations of the study, and future directions. © 2025 IEEE.
Exploring the Potential of SWNT /M072V30Thin Films for In-Materio Reservoir Computing
(Institute of Electrical and Electronics Engineers Inc., 2025) Harshita Rai; Kshitij RB Singh; Deep Banerjee; Yuki Usami; Somenath Garai; Hirofumi Tanaka; Shyam S. Pandey
The exponential growth of data-driven technologies and a paradigm shift towards unconventional computing frameworks like reservoir computing (RC) have led to the search for novel materials and architecture manifesting high-dimensional non-linear dynamics. This study explored a multiple redox site bearing Keplerate material towards their suitability in RC systems. To accomplish this, Keplerate-decorated single-walled carbon nanotube composite thin film was utilized as an effective material system. The resulting composite depicts non-linear 1-V dynamics with hysteresis and higher harmonic generations, proving its mettle for further applications as in-materio RC devices. © 2025 International Society of Functional Thin Film Materials Devices (FTFMD).
Fault Detection in PV Grid Integrated System via Machine Learning Technology
(Institute of Electrical and Electronics Engineers Inc., 2025) Souvik Mitra; K. A. Chinmaya
This research presents a study on machine learning algorithms for detecting failures in solar photovoltaic (PV) integrated systems with an exclusive focus on electrical faults. As the use of solar energy increases, maintaining the dependability of PV systems becomes crucial. Using a dataset from a grid-connected PV system, several machine learning algorithms are investigated to find flaws, including Logistic Regression, Decision Tree, Naive Bayes, and Random Forest. The issue identification process is automated with the suggested technique, improving system performance, reducing maintenance costs, and increasing efficiency. The results show that machine learning can significantly enhance the sustainability and dependability of solar PV systems. © 2025 IEEE.