Early detection of eye diseases such as diabetic retinopathy, glaucoma, and macular degeneration is critical for preventing vision loss and to ensure timely diagnosis. Centralized training of deep neural networks for medical image analysis is often constrained by data privacy regulations and institutional boundaries. Federated learning (FL) allows model training to be decentralized without exchange of raw data. This research is a structured comparison of pretrained convolutional neural networks (CNN) AlexNet, VGG16, ResNet50, and EfficientNet-B7 with a vision transformer (ViT) in centralized and federated training schemes. The effects of heterogeneous and non-identically distributed (non-IID) client data on architectural stability remains unaddressed in ophthalmic classification. Non-IID client distributions are simulated using Dirichlet-based partitioning to model realistic hospital-level label skew. This study focuses on examining the response of various architectures to decentralized optimization when the client distribution is non-uniform, focusing in particular on the stability of global models across training rounds. The sensitivity of global models to non-IID label skew and the federated aggregation rate at which global models converge to the correct solution are the main challenges addressed in this work. The results demonstrate substantial variability in architectural robustness under non-IID federated conditions, indicating that model depth and structural complexity significantly influence decentralized convergence. These are supplemented to the practical knowledge of federated learning in classification of ophthalmic images.

The advent of DNA microarray datasets has revolutionized both machine learning (ML) and healthcare by providing critical insights into variations in gene expression across different tissue and cell samples. These datasets, which typically contain thousands to tens of thousands of features, are instrumental in aiding disease diagnosis and classification. However, many genes within these datasets are irrelevant or noisy, making classification tasks challenging due to the high dimensionality of the data and the limited sample sizes. Feature selection is a crucial strategy to mitigate these challenges, as it helps identify the most relevant genes while discarding the irrelevant ones. This process not only improves the performance of ML models but also plays a pivotal role in accurately classifying tumor types and refining treatment strategies. This chapter reviews recent advancements in feature selection methods specifically tailored for microarray data analysis. It highlights the common challenges, including data imbalance, complexity, and missing values, which can hinder the classification process. Additionally, it provides an empirical analysis of various feature selection techniques applied to representative datasets, offering a comparative perspective rather than identifying a single “best” method. Importantly, the interpretation of gene signatures resulting from feature selection also has potential applications in computational social science, such as identifying key factors influencing public health, social behaviors, and policy decisions.

Multimedia storage and image sharing over cloud platforms have increased the need for secure and privacy-preserving image authentication. Traditional watermarking methods either operate in the plaintext domain or rely on Fully Homomorphic Encryption (FHE), which incurs high computational cost and limits real-time applicability. To address these issues, this paper proposes an efficient watermarking framework that integrates Lightweight Homomorphic Encryption (LHE) with the Discrete Wavelet Transform (DWT), enabling watermark embedding and extraction directly in the encrypted domain without decryption. The host image is encrypted using LHE and decomposed using DWT, where the watermark is embedded in the low-frequency LL sub-band to ensure high imperceptibility and robustness against common attacks. Compared to FHE- based methods, LHE significantly reduces computational time and memory usage. Experimental results on grayscale images demonstrate high visual quality in terms of PSNR and SSIM, along with reliable watermark extraction. The proposed method is suitable for privacy-sensitive applications such as medical imaging, cloud storage, and digital rights management.

Ensemble learning improves predictive performance by combining multiple classifiers; however, indiscriminate aggregation introduces redundancy, computational cost, and limited interpretability. This paper proposes a compact stacked ensemble framework that selectively retains classifiers exhibiting reliable probabilistic behavior. A γ′-pruning model based on fuzzy-rough set theory evaluates classifier relevance using out-of-fold posterior probability estimates, Gaussian similarity relations, and a fuzzy-rough dependency measure to capture probabilistic consistency beyond accuracy-based criteria. A greedy selection strategy balances validation performance with probabilistic stability. Experiments on medical and bioinformatics benchmarks demonstrate significant ensemble size reduction without sacrificing predictive performance, supporting fuzzy-rough probabilistic relevance as an effective criterion for classifier selection in uncertainty-aware stacked ensemble systems.

The computer-aided system to be used to detect Alzheimer disease (AD) should have computationally efficient models that are resistant to distortions that usually occur in medical imaging. This work presents a lightweight model of deep learning that is robustness-based and has the ability to classify AD by stage using structural MRI data. The given framework is based on the EfficientNetB0 backbone because it has a compound scaling property to enable efficient features learning with lower training complexity. The label-conscious augmentation and adversarial learning through the Fast Gradient Sign Method (FGSM) is also optimized to give a better resistance to imperceptible perturbation. Experimental assessment illustrates a test accuracy of 98.36% percent and an adversarial accuracy of 99.43% under Fast Gradient Sign Method (FGSM)-based adversarial evaluation through four stages of AD, namely, Non-Demented, Very Mild-Demented, Mild-Demented and Moderate-Demented. These findings suggest that the proposed solution is capable of high generalization, minimal computation cost, and high resistance to adversarial changes and would be an effective solution to clinical decision support in a resource-limited healthcare environment.

Reliable biomedical image classification is essential for computer-aided diagnosis systems, where incorrect or overconfident predictions may lead to serious clinical consequences. Although deep convolutional neural networks (CNNs) have achieved impressive performance in medical image analysis, individual models often suffer from limited robustness under class imbalance, inter-class similarity, and uncertainty caused by imaging artifacts. Ensemble learning partially mitigates these issues; however, most existing ensemble strategies rely on static fusion rules that fail to account for sample-wise model reliability. In this paper, we present an UncertaintyAware Fuzzy Ensemble (UAFE) framework for biomedical image classification. The proposed approach dynamically assigns voting weights to heterogeneous CNN models using calibrated confidence and entropy- based uncertainty within a fuzzy inference system that improves decision stability and interpretability by emphasizing reliable predictions while suppressing uncertain ones. The framework is evaluated on two clinically significant and complementary datasets i.e. a Chest X-ray dataset for pulmonary disease classification and a Brain MRI dataset for neurological disorder classification. Experimental results demonstrate that the proposed UAFE framework consistently outperforms individual CNNs and conventional voting-based ensembles in terms of accuracy, F1score, and prediction confidence. These findings highlight the effectiveness of confidence-driven fuzzy ensemble learning for robust and trustworthy medical image analysis.

High-dimensional datasets often contain redundant and highly correlated features that increase computational complexity and reduce model interpretability, making effective redundancy-aware feature selection an important challenge for fuzzy–rough learning. This paper proposes a Kohonen Self-Organizing Map-guided Fuzzy-Rough Feature Selection (KFRFS) method that enhances dependency-based fuzzy–rough feature selection by explicitly balancing feature relevance and structural redundancy during reduct construction. The proposed approach first organizes conditional attributes into topology-preserving groups using a Kohonen Self-Organizing Map trained on the transposed decision system and then integrates fuzzy–rough dependency gain with a KSOM-guided redundancy penalty through a unified scoring function within a greedy forward search framework. Extensive experiments on 17 benchmark datasets, including low-, medium-, and high-dimensional microarray datasets, demonstrate that KFRFS consistently produces compact approximate reducts while maintaining or improving classification performance compared with several state-of-the-art fuzzy–rough feature selection methods. Parameter sensitivity, ablation, cumulative dependency, statistical significance, and ranking analyses further show that the proposed redundancy-control mechanism provides a practical trade-off between dependency maximization and redundancy suppression, yielding a balanced relationship between reduct compactness and predictive accuracy. These results suggest that KFRFS can serve as a topology-aware extension of dependency-based fuzzy–rough feature selection for obtaining compact feature subsets in high-dimensional data analysis.

Gene selection is a critical task in analyzing high-dimensional DNA microarray data, where a vast number of genes are measured against a limited number of samples. Identifying the most informative subset of genes not only enhances classification accuracy but also reduces computational complexity, particularly in applications like cancer diagnosis and tumor classification. Fuzzy–rough set theory is an effective method for feature selection, capable of efficiently handling data imprecision and uncertainty during the selection process. This paper proposes a fuzzy–rough set-based gene-selection method (FRGS) that combines three steps: an early-accept forward selection, positive-region removal, and a backward elimination phase. Together, these steps limit redundant evaluations and remove unnecessary genes while relying on the fuzzy–rough dependency measure to guide the search. FRGS is evaluated on nine benchmark microarray datasets and compared with existing state-of-the-art methods. On every dataset, FRGS shortens selection time and preserves or slightly improves classification accuracy. Because the algorithm works directly on raw expression values and requires no discretizations or specialized parameter tuning, it can be applied readily in laboratory settings and large-scale gene-expression studies.

The hybrid paradigm of fuzzy-rough set theory, which combines fuzzy and rough sets, has proven effective in attribute reduction for hybrid decision systems encompassing both numerical and categorical attributes. However, current parallel/distributed approaches are limited to handling datasets with either categorical or numerical attributes and often rely on fuzzy dependency measures. There exists little research on parallel/distributed attribute reduction for large-scale hybrid decision systems. The challenge of handling high-dimensional data in hybrid decision systems necessitates efficient distributed computing techniques to ensure scalability and performance. MapReduce, a widely used framework for distributed processing, provides an organized approach to handling large-scale data. Despite its potential, there is a noticeable lack of attribute reduction techniques that leverage MapReduce's capabilities with a fuzzy discernibility matrix, which can significantly improve the efficiency of processing high-dimensional hybrid datasets. This paper introduces a vertically partitioned fuzzy discernibility matrix within the MapReduce computation model to address the high dimensionality of hybrid datasets. The proposed MapReduce strategy for attribute reduction minimizes data movement during the shuffle and sort phase, overcoming limitations present in existing approaches. Furthermore, the method's efficiency is enhanced by integrating a feature known as SAT-region removal, which removes matrix entries that satisfy the maximum satisfiability conditions during the attribute reduction process. Extensive experimental analysis validates the proposed method, demonstrating its superior performance compared to recent parallel/distributed methods in attribute reduction.

Vehicle-to-Vehicle (V2V) communication plays a major role in smart transportation systems for autonomous and connected vehicles, which requires confidentiality and data integrity. Data sharing occurs over an insecure channel using conventional cryptography, which is prone to quantum attacks. Therefore, V2V needs a secure authentication scheme. This model proposes a lightweight quantum-resistant V2V authentication scheme that integrates lattice signatures with a superdense coding quantum communication channel through the blockchain. The proposed model uses blockchain technology to securely store and share lattice signatures and keys. For secure communication, authentication is performed between vehicles and roadside units using the Dilithium lattice signature through a decentralized system for data integrity and confidentiality. The experimental analysis states that using the authentication scheme, the proposed model has low latency between V2V and RSU. The security analysis of the proposed model is performed through formal security verification using the Scyther tool and informal security analysis. The security analysis proves that the proposed protocol is resistant to well-known attacks.

Fuzzy-rough set theory is an efficient method for attribute reduction. It can effectively handle the imprecision and uncertainty of the data in the attribute reduction. Despite its efficacy, current approaches to fuzzy-rough attribute reduction are not efficient for the processing of large data sets due to the requirement of higher space complexities. A limited number of accelerators and parallel/distributed approaches have been proposed for fuzzy-rough attribute reduction in large data sets. However, all of these approaches are dependency measure based methods in which fuzzy similarity matrices are used for performing attribute reduction. Alternative discernibility matrix based attribute reduction methods are found to have less space requirements and more amicable to parallelization in building parallel/distributed algorithms. This paper therefore introduces a fuzzy discernibility matrix-based attribute reduction accelerator (DARA) to accelerate the attribute reduction. DARA is used to build a sequential approach and the corresponding parallel/distributed approach for attribute reduction in large data sets. The proposed approaches are compared to the existing state-of-the-art approaches with a systematic experimental analysis to assess computational efficiency. The experimental study, along with theoretical validation, shows that the proposed approaches are effective and perform better than the current approaches.

The scale of the data collected today from applications in the real-world is massive. Sometimes this data can also include missing (incomplete) values that give rise to large-scale incomplete decision systems (IDS). Parallel attribute reduction in big data is an essential preprocessing step for scalable machine learning model construction. Rough set theory has been used as a powerful tool for attribute reduction in complete decision systems (CDS). Furthermore extensions to classical rough set theory have been proposed to deal with IDS. A lot of research works have been done on efficient attribute reduction in IDS using these extensions, but no parallel/distributed approaches have been proposed for attribute reduction in large-scale IDS. Since, owing to its two challenges, large-scale and incompleteness, the processing of large-scale IDS is difficult. To address these challenges, we propose MapReduce based parallel/distributed approaches for attribute reduction in massive IDS. The proposed approaches resolve the challenge of incompleteness with the existing Novel Granular Framework (NGF). And each proposed approach follows a different data partitioning strategy to handle the data sets that are large-scale in terms of number of objects and attributes. One of the proposed approaches adopts an alternative representation of the NGF and uses a horizontal partitioning (division in object space) of the data to the nodes of cluster. Another approach embraces the existing NGF and uses a vertical partitioning (division in attribute space) of the data. Extensive experimental analysis carried out on various data sets with different percentages of incompleteness in the data. The experimental results show that the horizontal partitioning based approach performs well for the massive object space data sets. And the vertical partitioning based approach is relevant and scales well for extremely high dimensional data sets.

In the last few decades, rough sets have evolved to become an essential technology for feature subset selection by way of reduct computation in categorical decision systems. In recent years with the proliferation of MapReduce for distributed/parallel algorithms, several scalable reduct computation algorithms have been developed in this field for large-scale decision systems using MapReduce. The existing MapReduce based reduct computation approaches use horizontal partitioning (division in object space) of the dataset into the nodes of the cluster, requiring a complicated shuffle and sort phase. In this work, we propose an algorithm MR_IQRA_VP which is designed using vertical partitioning (division in attribute space) of the dataset with a simplified shuffle and sort phase of the MapReduce framework. MR_IQRA_VP is a distributed/parallel implementation of the Improved Quick Reduct Algorithm (IQRA_IG) and is implemented using iterative MapReduce framework of Apache Spark. We have done an extensive comparative study through experimentation on benchmark decision systems using existing horizontal partitioning based reduct computation algorithms. Through experimental analysis, along with theoretical validation, we have established that MR_IQRA_VP is suitable and scalable to datasets of larger size attribute space and moderate object space prevalent in the areas of Bioinformatics and Web mining.

Fuzzy-rough set theory (FRST) is a hybridization of fuzzy sets with rough sets with applications to attribute reduction in hybrid decision systems. The existing reduct computation approaches in fuzzy-rough sets are not scalable to large scale decision systems owing to higher space complexity requirements. Iterative MapReduce framework of Apache Spark facilitates the development of scalable distributed algorithms with fault tolerance. This work introduces algorithm MR-FRDM-SBE as one of the first attempts towards scalable fuzzy-rough set based attribute reduction. MR-FRDM-SBE algorithm is a combination of a novel incremental approach for the construction of distributed fuzzy-rough discernibility matrix and Sequential Backward Elimination control strategy based distributed fuzzy-rough attribute reduction using a discernibility matrix. A comparative experimental study conducted using large scale benchmark hybrid decision systems demonstrated the relevance of the proposed approach in scalable attribute reduction and better classification model construction.

Fuzzy-rough set theory, an extension to classical rough set theory, is effectively used for attribute reduction in hybrid decision systems. However, it’s applicability is restricted to smaller size datasets because of higher space and time complexities. In this work, an algorithm MR_IMQRA is developed as a MapReduce based distributed/parallel approach for standalone fuzzy-rough attribute reduction algorithm IMQRA. This algorithm uses a vertical partitioning technique to distribute the input data in the cluster environment of the MapReduce framework. Owing to the vertical partitioning, the proposed algorithm is scalable in attribute space and is relevant for scalable attribute reduction in the areas of Bioinformatics and document classification. This technique reduces the complexity of movement of data in shuffle and sort phase of MapReduce framework. A comparative and performance analysis is conducted on larger attribute space (high dimensional) hybrid decision systems. The comparative experimental results demonstrated that the proposed MR_IMQRA algorithm obtained good sizeup/speedup measures and induced classifiers achieving better classification accuracy.