Browse Items (16481 total)
Sort by:
-
Feature selection based on the classifier models: Performance issues in the pre-diagnosis of lung cancer /
Journal of Theoretical and Applied Information Technology, Vol-59(3), pp.549-555. ISSN-1992-8645. -
Feature selection based on the classifier models: Performance issues in the prediagnosis of lung cancer
Dimensionality reduction is generally carried out to reduce the complexity of the computations in the large data set environment by removing redundant or de-pendent attributes. For the Lung cancer disease prediction, in the pre-diagnosis stage, symptoms and risk factors are the main information carriers. Large number of symptoms and risk attributes poses major challenge in the computation. Here in this study an attempt is made to compare the performance of the attribute selection models prior and after applying the classifier models. A total of 16 classifier models are preferred based on relevancy of the models with respect to the data types chosen, which are based on statistical, rule based, logic based and artificial neural network approaches. Feature set selection and ranking of attributes are done based on individual models. Based on the confusion matrix parameters the models prediction outcomes are found out in the supervisory training mode. The Confusion matrix of the models before and after dimensionality reduction is computed. Models are compared based on weighted Reader Operator Characteristics. Normalized weights are assigned based for the result of individual models and predictive model is developed. Predictive models performance is studied with target under supervised classifier model and it is observed that it is tallying with the expected outcome. 2005 - 2014 JATIT & LLS. All rights reserved. -
Feature selection/dimensionality reduction
In today's world, medical image analysis is a critical component of research, and it has been extensively explored over the last few decades. Machine learning in healthcare is a fantastic advancement that will improve disease detection efficiency and accuracy. In many circumstances, it will also allow for early detection and treatment in remote or developing areas. The amount of medical data created by various applications is growing all the time, creating a bottleneck for analysis and necessitating the use of a machine learning method for feature selection and dimensionality reduction techniques. Feature selection is an important concept of machine learning since it affects the model's performance and the data parameters you utilize to train your machine learning models to have a big influence on the performance. The approach of minimizing the number of inputs in training data by reducing the dimension of your feature set is known as dimensionality reduction. Reduced dimensionality aids in the overall performance of the machine learning algorithms. 2023 River Publishers. -
Feature Subset Selection Techniques with Machine Learning
Scientists and analysts of machine learning and data mining have a problem when it comes to high-dimensional data processing. Variable selection is an excellent method to address this issue. It removes unnecessary and repetitive data, reduces computation time, improves learning accuracy, and makes the learning strategy or data easier to comprehend. This chapterdescribes various commonly used variable selection evaluation metrics before surveying supervised, unsupervised and semi-supervised variable selection techniques that tend to be often employed in machine learningtasks including classification and clustering. Finally, ensuing variable selection difficulties are addressed. Variant selection is an essential topic in machine learning and pattern recognition, and numerous methods have been suggested. This chapter scrutinizesthe performance of various variable selection techniques utilizing public domain datasets. We assessed the quantity of decreased variants and the increase in learning assessment with the selected variable selection techniques and then evaluated and compared each approach based on these measures. The evaluation criteria for the filter model are critical. Meanwhile, the embedded model selects variations during the learning model's training process, and the variable selection result is automatically outputted when the training process is concluded. While the sum of squares of residuals in regression coefficients is less than a constant, Lasso minimizes the sum of squares of residuals, resulting in rigorous regression coefficients. The variables are then trimmed using the AIC and BIC criteria, resulting in a dimension reduction. Lasso-dependent variable selection strategies, such as the Lasso in the regression model and others, provide a high level of stability. Lasso techniques are prone to high computing costs or overfitting difficulties when dealing with high-dimensional data. 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG. -
Features of Vitamin Model Affecting Psychological Empowerment: Serial Mediation Role of Job Crafting and Work Engagement
The current research aimed to investigate the association between the variables under the study, that is, the vitamin model features of a job: job crafting, work engagement, and psychological empowerment. It also attempted to analyze the serial mediational role of the two causally linked mediators, that is, job crafting and work engagement with the job features of the vitamin model and psychological empowerment. By investigating these variables, we tried to explore how the employees redesigned the well-defined jobs to match their capabilities, which enhanced commitment to work and led to positive behavioral outcomes, such as empowerment, work meaningfulness, and improved performance. Primary data were collected from 453 knowledge workers in the information technology (IT) and information technology-enabled services (ITES) industry. Using SPSS software, the correlation method revealed significant positive correlations between the variables under study. PROCESS macro (Haynes, 2012) was applied in SPSS AMOS regression-based path coefficients and bootstrap confidence intervals at a 95% confidence level. As the bootstrap confidence intervals did not include zero, a significant mediational role of the serial mediators was observed between the relationship of features in the vitamin model and psychological empowerment [Estimate =.0761, 95% CI (.0257,.1902)]. So, it could be concluded that job crafting made the employees the mechanic of their vehicle (work), leading to work engagement, increased performance, and psychological well-being at the workplace. 2022, Associated Management Consultants Pvt. Ltd.. All rights reserved. -
Featuring Machine Learning Models to Evaluate Employee Attrition: A Comparative Analysis of Workforce Stability Relating Factors
Employee attrition is a problem for most organizations as it affects morale, productivity, and business continuity. In addressing this, the study made use of machine learning techniques such as Clear AI, Random Forest, and logistic regression in designing a prediction model to predict who is the next to leave within an organization. The HR data relating to demographics, performance metrics, job roles, and conditions of work was sourced from publicly available website Kaggle.com for the study. Data preprocessing included scaling, outlier detection, and balancing the dataset using SMOTE. Multiple machine learning models were trained and evaluated by checking on accuracy, F1-score, and the ROC-AUC curve. The best model that was tested was Random Forest, which gave an accuracy of 85.71%. Additional insights from feature importance highlighted the significant effect of overtime, marital status, and stock options on attrition. Among the remaining key drivers are workload, work-life balance, and financial incentives. These findings suggest the need for focused HR strategies, such as reduction of overtime, mentorship programs, and career development opportunities, to reduce attrition rates and improve employee satisfaction. This study provides a robust methodology in predicting attrition and delivers actionable insights into designing interventions that improve workforce stability and organizational efficiency. 2025, Iquz Galaxy Publisher. All rights reserved. -
FEC & BCH: Study and implementation on VHDL
Channel encoding and Forward Error Correction is a crucial element of any communication system. This paper gives a brief overview of the fundamentals, mechanism and importance of Forward Error Correction. The design and implementation of a (63,36,5) BCH Codec is also projected in the later sections. All simulations are made on MATLAB R2018b and the VHDL implementations have been carried out using Xilinx Vivado 2018.2. 2019 IEEE -
FeCl3/KOH two steps activated biocarbon with hierarchical porosity and oxygen-rich for enhanced supercapacitor applications
Biomass waste derived from jackfruit (Artocarpus heterophyllus) cores is used to fabricate hierarchical porous activated carbon through chemical activation with Iron(III) chloride (FeCl3) and potassium hydroxide (KOH). Jackfruit is an abundant agricultural by-product in tropical regions, including India, Bangladesh, and Sri Lanka. The activated carbon derived from jackfruit provides a sustainable, low-cost, and high-performance alternative to conventional carbon materials for supercapacitors, thereby aligning with waste valorisation strategies. The prepared carbon displays hierarchical porous structures of both micro and mesopore architectures. They are amorphous and contain functional oxygen groups, as confirmed by X-Ray photoelectron spectroscopy (XPS) and Fourier Transform Infrared Spectroscopy (FTIR). A high surface area (1251m2g?1) was obtained via Brunauer-Emmett-Teller (BET) analysis. The electrochemical performances, via cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and galvanostatic charge/discharge (GCD) show high specific capacitance of 310Fg?1 at 0.8Ag?1 from GCD, 331Fg?1 at 10mVs?1 from CV, and a charge transfer resistance of 0.1410?cm2, in three electrode configuration and showing good cycling stability of 87% over 2500 cycles. These results suggest that the activated carbon offers potential application in low-cost and renewable production of carbon materials for supercapacitors applications. 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies. -
FedDiff-Health: A Privacy-Preserving Generative Framework for Collaborative Hospital Readmission Prediction
Hospital readmission prediction encounters three challenges: data siloing across hospitals due to format incompatibilities, stringent privacy constraints, and the rarity of readmission events. We propose P-Fed-Diffusion, the first framework that enables collaboration across hospitals while keeping patients' data private. Our method automatically aligns heterogeneous data schemas without human intervention, using large language models. Then, we apply conditional diffusion models within a federated learning framework to generate synthetic data for rare readmission events. The framework incorporates formal privacy guarantees via differential privacy. We achieve a dramatic improvement over state-of-the-art methods: while the best prior method achieves 2% recall, we achieve 64% recall-32x improvement, meaning that the method finds over 1,000 additional high-risk patients per hospital annually. Our work opens up a new direction for privacy-preserving collaborative AI across hospitals. 2025 IEEE. -
Federated and Explainable AI Models for Secure FinTech Transactions in Digital Manufacturing Supply Chains
Digital manufacturing supply chains are becoming increasingly dependent on inbuilt FinTech services to perform automated payments, invoicing, and settlements which presents sensitive financial and operational data to security and privacy threats. This article is an empirical paper concerning the application of Federated Learning (FL) and Explainable Artificial Intelligence (XAI) in securing FinTech transactions in decentralized manufacturing supply chains. The suggested framework will facilitate joint fraud and anomaly-related detection without exchanging raw data between supply-chain participants. Different privacy mechanisms such as client-level and secure aggregation are integrated to safeguard sensitive data and minimize the risks of inferences. Explainable AI methods are used such as SHAP, local surrogate models, to enable transparency and auditability as well as regulatory compliance. Experimental evidence has shown that federated models can attain almost centralized detection accuracy with much stronger privacy guarantees and explainability procedures can give insightful and interpretable information about model decisions. The paper identifies the trade-offs between accuracy, privacy, and computational overhead and concludes that federated and explainable AI provides a convenient, secure, and compliant solution to FinTech-enabled digital manufacturing ecosystems. 2026 IEEE. -
Federated Learning and Blockchain: A Cross-Domain Convergence
Gaining significant attention within decentralized contexts, Federated Learning (FL) has been positioned as a highly desirable method for machine learning. By enabling multiple entities to train a shared model cooperatively, data privacy and security are preserved by Federated Learning. Harnessing inherent transparency and accountability of blockchain technology to trace and authenticate updates effectively in federated learning has transpired as an up-and-coming avenue to tackle data challenges related to confidentiality, protection, and reliability. This study examines the viability of federated learning and blockchain integration across multiple dimensions. The technological components of this integration., including incentive systems, consensus mechanisms, data validation, and smart contracts, are delved into. In the study, a novel proposed model for federated learning integrated with blockchain is designed and implemented. It is observed that the mean cypher size is 100 bytes for varying values of gradients. The average throughput recorded is 1.7 bytes per second, while the mean accuracy is 87.1% for 50 epochs. 2023 IEEE. -
Federated Learning for Privacy-Preserving Threat Detection in IoT-Enabled Networks
The advent of Internet of Things (IoT) devices at a continuous rapid pace has greatly increased the surface for cyberattacks to measure the effectiveness of threat detection mechanisms. Most conventional centralized threat detection frameworks require sending sensitive device data to a single central server for aggregation, with significant privacy risks and scalability challenges. Such challenges could be efficiently addressed with the use of Federated Learning (FL), an emerging decentralized paradigm of training machine learning models, through the collaboration of a large number of devices, such as IoT sensors, that store the data locally and do not share raw data. In this work, we integrate FL to propose a threat detection framework for preserving privacy in IoT-enabled networks. In this paper, we propose a system architecture in which edge devices perform local training of machine learning models on encrypted traffic and behavioral data and then periodically share only the model updates with a centralized aggregator. This approach ensures the privacy of the data, minimizes communication overhead, and improves detection capabilities for real-time threats. The efficacy of FL-based threat detection is examined through experimental evaluations on benchmark datasets of IoT attack traces, indicating that FL-based approaches achieve competitive accuracy versus prior centralized schemes while greatly mitigating risks of data leakage. We further address issues regarding heterogeneous device resources, communication efficiency, and adversarial attack resilience in this context. Our results indicate that federated learning is a very effective approach for providing IoT environment protection, as it securely balances privacy, scalability, and detection performance. 2025 IEEE. -
Federated Learning with Adaptive Intermediate Model Selection for Predicting IVIG Resistance in Kawasaki Disease
Kawasaki disease (KD), a rare pediatric illness affecting children under five, is treated with intravenous immunoglobulin (IVIG). But 1020% of patients are resistant to IVIG, and these resistant kids face a higher risk of coronary artery abnormalities. Identifying resistance early is vital, yet data scarcity, class imbalance, and the diseases rarity necessitate nationwide collaboration, which is often hindered by country-specific privacy policies. Federated learning (FL) provides a practical way for different parties to collaborate on training a model while keeping their raw data private and secure. To enhance model adaptability across diverse clinical populations, we propose an adaptive intermediate model selection strategy in federated learning. Each client retains the versionglobal or locally fine-tunedthat performs best on its own data, using customizable performance metrics such as F1-score or recall. The system was implemented using the Flower FL framework, with three simulated clients and a shared convolutional neural network (CNN) architecture. Experiments demonstrated that the global model achieved stronger performance than conventional models, and several clients obtained further gains by selecting intermediate models aligned with their data. This approach introduces a novel balance between worldwide collaboration and local personalization in FL, offering a flexible and clinically meaningful solution for IVIG resistance prediction. 2026 by the authors of this article. -
Federated Multi-Task Learning Using Server-Side Normalized Loss-Based Weighting Method
Federated multi-task learning is an approach where multiple clients collaboratively train related but distinct models on their local data without sharing it, thereby preserving privacy while leveraging collective knowledge. However, participating clients can have very different data distributions, sizes and quality, leading to statistical heterogeneity. This heterogeneity is a major challenge in federated learning, as noisy or inconsistent updates from some clients can slow down convergence or degrade the global model's performance. MOCHA is a seminal federated multi-task learning framework that explicitly models task relationships and optimizes clientspecific models, while addressing system challenges like communication costs, fault tolerance and client dropouts. In this work, we enhance MOCHA with a server-side normalized lossbased weighting technique that focuses on the quality of client updates. Each client in the federated multi-task setup computes its local training loss, which is sent to the server during communication rounds. The server normalizes these losses across clients and assigns adaptive aggregation weights, giving more influence to clients with lower normalized losses and down-weighting noisy or unreliable clients. This design simplifies client-side implementation because all weighting is performed at the server. Experiments on heterogeneous MNIST and CIFAR-10 tasks show that the proposed method achieves a slightly higher final-round average test accuracy (0.5108 vs. 0.5065), reduces average training loss by approximately 2.6% (from 1.1148 to 1.0858), and improves fairness by lowering the standard deviation of client accuracies by about 5% (from 0.3631 to 0.3450) compared to baseline MOCHA. These results indicate that server-side normalized loss-based weighting improves training stability, convergence behavior and crossclient fairness in federated multi-task learning under nonconvex optimization. 2025 IEEE. -
FEDGE: FEDerated learning at the EDGE on space platforms using deep neural network architectures
We introduce FEDGE: FEDerated Learning at the EDGE, a framework designed for efficient AI deployment in resource-constrained satellite constellations. FEDGE integrates federated learning with edge computing to address communication overhead and latency challenges in distributed space environments. The framework features a novel edge-enhanced ground station protocol that dynamically schedules model aggregation based on satellite-provided metadata, combined with local stochastic gradient descent training at satellite edge devices and gradient compression via quantization. Experimental validation on MNIST and EuroSAT datasets demonstrates the practical viability of the approach. On MNIST, FEDGE achieved 94.33% training accuracy with 0.21 loss and 90.05% test accuracy with 0.24 loss. On EuroSAT, the framework reached 93.47% training accuracy with 0.18 loss and 91.51% test accuracy with 0.21 loss. Gradient quantization reduces data exchange by up to 14 with approximately 4% impact on test loss. These results validate FEDGE as a communication-efficient solution for decentralized AI deployment in satellite systems, enabling autonomous spacecraft intelligence and addressing the unique constraints of space-based computing platforms. The Author(s) 2025. -
Feedstocks for production of polyhydroxyalkanoates: Sugar-and starch-rich waste as fermentation substrates
Effective management of food and agricultural waste faces a crucial challenge in today's world, primarily due to the necessity to sustainably feed the ever-growing global population. A signifcant portion of this waste generated is plant waste, including both sugar-rich and starch-rich materials. These waste materials are often considered as non-product leftovers due to their perceived lack of economic value compared to the costs associated with their collection, storage, and recovery for reuse. However, by employing appropriate technological methods, such wastes can be utilized as feedstocks for the production of value-added products such as polyhydroxyalkanoates (PHA). PHAs are biodegradable polymers with a wide range of applications, offering an environmentally friendly potential substitute to the conventional plastics. Recycling plant wastes holds immense potential for application across various industries, ultimately leading to a reduction in the adverse impacts caused by their accumulation in the environment. This chapter delves into the utilization of plant wastes (sugar-rich and starch-rich wastes including beet molasses, sugarcane molasses, corn steep liquor, and starchy wastewater) for the production of PHAs. It discusses PHA recovery methods and characterization techniques crucial for evaluating the properties of PHA, thereby laying the groundwork for understanding the material quality and suitability for various applications. Additionally, diverse applications of PHAs, ranging from packaging materials to biomedical devices, are explored by highlighting the potential of utilizing plant wastes to contribute to a circular economy. Springer Nature Singapore Pte Ltd. 2025. All rights reserved. -
Female Director and Agency Cost: Does board gender diversity at Indian corporate board reduce agency conflict?
We examined the presence of women directors in top-level management and their effect on principal-principal conflict (PP) and principal-agent conflict (PA) on the firms listed on Indian stock exchange using a panel model approach. For analysis purpose, this study covers the sample of 75 companies belonging to various industries and listed in Bombay Stock Exchange Index, has been studied over thirteen financial years, i.e. from year 2006 to year 2019. This study uses panel data analysis, i.e. fixed effect model and random effect model. The proportion and presence (dichotomous) of women directors on top level management board is taken as the independent variable. Principalprincipal conflict measured by assets utilization ratio (AUR), and principal-agent conflict is been measured by dividend payout ratio (DPR), are taken as dependent variable in this study. The prime results of this study using panel data analysis, i.e. fixed effect (FE) and random effects (RE) estimation models point towards no significant impact of the female director (proportion and presence) on the firm's agency cost (PP and PA). 2021. Transnational Press London. All Rights Reserved. -
Female entrepreneurship: Challenges faced in a global perspective
Women being employees is a very appreciative aspect. But women being employers is a bold and massive decision that they make, considering their busy life schedule, which traditionally includes looking after their families and themselves. This book chapter aims to identify the diverse challenges female entrepreneurs face, which could be in the context of society, structure, or finance. Identifying and vocalizing these challenges faced by these emerging entrepreneurs is inadequate, but tackling them is equally called for. This study provides a framework and scope for further research to look into more opportunities and measures to tackle these roadblocks. Last but not least, this book chapter anticipates inspiring all women and driving them to explore the essence of entrepreneurship. 2023, IGI Global. All rights reserved. -
Female masculinities and women of third nature: Analyzing the gender and sexual politics of identity and visibility of alternative masculinities through indian mythologies and literary narratives
Alternative sexualities have been a part of the Indian past since time immemorial, and mention of them is often visible in Sanskrit mythological texts. As much as the presence of hijras and other gendered cultural identities is in Indian and Western public discourses, there is a narrow space occupied by women of the third kind with female masculinities, with scant attention leading to the higher invisibility of women of the third kind. Female masculinity is often considered a "rejected shred, " while male masculinity is seen as real and heroic. This chapter focuses on "masculinity without men" to explore alternative masculinities-the concept popularized by Judith Halberstam (Judith Halberstam, Female Masculinity. Zubaan Books, New Delhi, 1998). We delve into the politics of alternative modes of enactment and production where male masculinity is embedded. This chapter centers on female masculinity and alternative forms of masculinities performed, enacted, and embodied by female individuals as reflected in the Indian past and mythology. This chapter further delves into identifying histories and representations of female masculinities in Indian literature to bring female masculinities and women of the third kind into academic discourse. Springer Nature Switzerland AG 2022. All rights reserved. -
Female Political Representation and Economic Development in India: An Empirical Analysis
Recent years have seen an enhanced focus on women's roles in politics, with research increasingly showing that having a more significant gender representation in decision-making roles can significantly impact economic growth. This chapter delves into how women's political involvement, economic advancement, and gender equality have evolved in India over twenty years from 2000 to 2020, using a time series analysis. The study uses vector autoregression (VAR) analysis to examine how political representation of female, participation rate of labour force (LFPR), and health investment affect the Gender Development Index (GDI). The model diagnostics successfully demonstrated stationarity, non-serial correlation, and the lack of homoscedasticity. The analysis highlights that Female LFPR and GDI are positively related, whereas health expenditure and GDI are negative. Female labour market participation improves GDI, whereas females consistently receive less healthcare expenditure than males, leading to a negative relationship between health expenditure and GDI. Importantly, it is observed that labour market participation has a more substantial effect on GDI than political representation or health investments. This shows that greater female labour force participation is more critical in gender equality than increased political representation or healthcare spending. Highlighting the necessity for policies tailored to women, the chapter argues that these measures are critical for enhancing LFPR and boosting GDI and societal progress. The chapter contributes to the gender discourses in political participation and the empowerment of female, proposing a strategy to improve women's contribution to the labour market, leading higher GDI and, as a result, a more equitable society. 2026 selection and editorial matter, Hebatallah Adam and Abul Hasnat Monjurul Kabir; individual chapters, the contributors. All rights reserved.